Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Miniature phased-array transducer for colorectal tissue characterization during TEM robotic surgery; and, Forward-looking phased-array transducer for intravascular imaging
(USC Thesis Other)
Miniature phased-array transducer for colorectal tissue characterization during TEM robotic surgery; and, Forward-looking phased-array transducer for intravascular imaging
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
MINIATURE PHASED-ARRAY TRANSDUCER FOR COLORECTAL TISSUE CHARACTERIZATION DURING TEM ROBOTIC SURGERY AND FORWARD-LOOKING PHASED-ARRAY TRANSDUCER FOR INTRAVASCULAR IMAGING by Nestor Emmanuel Cabrera-Munoz A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (BIOMEDICAL ENGINEERING) May 2018 Copyright 2018 Nestor Emmanuel Cabrera-Munoz ii To my Family Thank you God for everyday blessings iii COMMITTEE MEMBERS K. Kirk Shung, Ph.D. (Chair) Dean’s Professor in Biomedical Engineering University of Southern California – University Park Campus Mike S.W. Chen, Ph.D. Assistant Professor of Electrical Engineering University of Southern California – University Park Campus Cristina Zavaleta, Ph.D. Gabilan Assistant Professor of Biomedical Engineering University of Southern California – University Park Campus iv ABSTRACT This dissertation describes the design, fabrication, and testing of two different high- frequency ultrasound array probes for medical imaging applications. State-of-the-art piezoelectric materials, fine spatial scale high-frequency composites, modeling tools, custom- engineered electrical and mechanical components, as well as novel fabrication techniques were employed to maximize the performance of the arrays. The first probe consists in a 15-MHz, side-looking, miniature linear phased array for colorectal robotic surgery. The array is intended to be integrated into a robotic surgery probe for the real-time intraoperative fusion of Endorectal Ultrasound (ERUS) and endomicroscopy to assist in the assessment of colorectal cancer surgical margins during Transanal Endoscopic Microsurgery (TEM) procedures. The measured performance and imaging evaluation confirmed that the fabricated array provides enough penetration (≥ 8.1 mm), peak-to-peak sensitivity (≥ 100 mV), and spatial resolution (90 μm axially / 420 μm laterally) for the overall structure detection of colorectal cancerous tissue. The second probe consists in a 30-MHz, forward-looking, miniature linear phased array for Intravascular Ultrasound (IVUS) imaging and intraluminal vessel navigation during Peripheral Arterial Disease (PAD) angioplasty procedures. The measured performance and imaging evaluation confirmed that the fabricated array provides enough depth of view (≥ 5.0 mm), peak-to-peak sensitivity (≥ 25 mV), and spatial resolution (65 μm axially / 215 μm laterally) for the overall detection of structures lying ahead of the catheter. v TABLE OF CONTENTS List of Figures ix List of Tables xxiii CHAPTER 1 Introduction 1 1.1 Medical Ultrasound 1 1.2 Ultrasound Principles 5 1.3 Ultrasound Transducers 15 1.4 Ultrasound Arrays 29 1.5 Objectives of Research 42 CHAPTER 2 Miniature Linear Phased Array for Colorectal Robotic Surgery 44 2.1 Clinical Problem: Colorectal Cancer (CRC) 44 2.2 Current Treatment: Transanal Endoscopic Microsurgery (TEM) 46 2.3 Clinical Need 50 2.4 Array Transducer Design 51 2.4.1 Array Acoustic Stack 53 2.4.2 Electrical Interconnect – Flexible Printed Circuit 56 2.4.3 Electrical Interconnect – Cabling 57 2.4.4 Electrical Interconnect – Verasonics Backshell Connector Kit 60 2.4.5 Verasonics Imaging System 61 2.5 Array Transducer Simulation Modeling 63 2.5.1 Pulse-Echo Response and Electrical Impedance 63 vi 2.5.1a Input File for Pulse Transmission 63 2.5.1b Input File for Echo Reception 68 2.5.1c Review File for Displaying Results 68 2.5.2 Electrical and Acoustical Crosstalk 70 2.5.2a Input File for Pulse Transmission on a Desired 70 Piezoelectric Element and for the Determination of Perceived Voltages on Four Adjacent Piezoelectric Elements 2.5.2b Review File for Displaying Results 71 2.5.3 Pressure Field and Axial/Lateral Beam Profiles 73 2.5.3a Input File for Pressure Wave Generation and Propagation 74 2.5.3b Review File for Displaying the Resulting Pressure 76 Field and Axial/Lateral Beam Profiles 2.6 Array Transducer Fabrication 79 2.6.1 Array Acoustic Stack (2-2 Composite + 1 st Matching Layer) 79 2.6.2 Flexible Printed Circuit 81 2.6.3 Bonding the Array Acoustic Stack and Backing Layer to the 82 Flexible Printed Circuit 2.6.4 Cable Assembly Connection to the Flexible Printed Circuit 85 2.6.4a Manual Micro-Soldering with Solder and Soldering Iron 86 2.6.4b Manual Cold-Soldering with Conductive Epoxy, 87 E-Solder 3022 2.7 Final Packaging and Connection to Verasonics Backshell Connector 91 2.8 Array Transducer Characterization 95 2.8.1 Electrical Impedance 95 2.8.2 Pulse-Echo Response and Insertion Loss 98 vii 2.8.3 Electrical and Acoustical Crosstalk 102 2.8.4 Single-Element Directivity Response 104 2.9 Array Transducer Imaging 105 2.9.1 Polished Quartz Reflector Imaging 106 2.9.2 Custom-Made Fine-Wire Phantom Imaging 108 2.9.3 Custom-Made Tissue-Mimicking Phantom with Needle Imaging 110 2.9.4 Custom-Made Tissue-Mimicking Phantom with Tubing Imaging 111 CHAPTER 3 Forward-Looking Intravascular Imaging Catheter 114 3.1 Clinical Problem: Peripheral Arterial Disease (PAD) 114 3.2 Current Treatment: Angioplasty 115 3.3 Clinical Need 121 3.4 Array Transducer Design 121 3.4.1 Array Acoustic Stack 123 3.4.2 Electrical Interconnect – Flexible Printed Circuit 125 3.4.3 Electrical Interconnect – Cabling 126 3.4.4 Electrical Interconnect – Verasonics Backshell Connector Kit 129 3.4.5 Verasonics Imaging System 131 3.5 Array Transducer Simulation Modeling 132 3.5.1 Pulse-Echo Response and Electrical Impedance 133 3.5.2 Electrical and Acoustical Crosstalk 134 3.5.3 Pressure Field and Axial/Lateral Beam Profiles 135 3.6 Array Transducer Fabrication 137 3.6.1 Array Acoustic Stack (2-2 Composite + 1 st Matching Layer) 138 viii 3.6.2 Flexible Printed Circuit 144 3.6.3 Bonding the Array Acoustic Stack and Backing Layer to the 144 Flexible Printed Circuit 3.6.4 Cable Assembly Connection to the Flexible Printed Circuit 148 3.7 Final Packaging and Connection to Verasonics Backshell Connector 152 3.8 Array Transducer Characterization 154 3.8.1 Electrical Impedance 155 3.8.2 Pulse-Echo Response and Insertion Loss 159 3.8.3 Electrical and Acoustical Crosstalk 163 3.8.4 Single-Element Directivity Response 164 3.9 Array Transducer Imaging 166 3.9.1 Polished Quartz Reflector Imaging 167 3.9.2 Custom-Made Fine-Wire Phantom Imaging 168 3.9.3 Custom-Made Porcine Carotid Artery Phantom Imaging 170 CHAPTER 4 Summary 174 4.1 Miniature Linear Phased Array for Colorectal Robotic Surgery 174 4.1.1 Array Development Summary 174 4.1.2 Array Performance Summary 175 4.2 Forward-Looking Intravascular Imaging Catheter 176 4.2.1 Array Development Summary 176 4.2.2 Array Performance Summary 177 Bibliography 180 ix LIST OF FIGURES Figure 1.1 2D/Color/Pulsed Wave Doppler combined mode image of the heart showing mitral valve regurgitation. (Courtesy of Peter Munk Cardiac Centre) 2 Figure 1.2 (A) GE Vivid S6 TM ultrasound system being used in a hospital setting. The GE Vivid S6 TM delivers imaging and analysis in a broad range of applications: Cardiac, vascular, abdominal, and pediatric/fetal (adapted from General Electric Healthcare, Inc.). (B) GE Vscan TM pocket-sized, handheld (dual probe version) ultrasound system and (C) GE Vscan TM (phased array single probe version) being used in a rural clinic. (Courtesy of General Electric Healthcare, Inc.) 3 Figure 1.3 2D color Doppler mode image of the umbilical cord showing real-time display of thermal (TI) and mechanical (MI) indices at the upper right corner. (Courtesy of Philips Healthcare, Inc.) 5 Figure 1.4 CW excitation of an elastic medium by an ultrasound source. 6 Figure 1.5 Particle motion versus the direction of wave propagation for longitudinal and shear waves. Longitudinal waves are the most commonly employed in medical ultrasound imaging. (Courtesy of Olympus, Inc.) 8 Figure 1.6 Reflection and transmission of a plane ultrasonic wave incident on an interface between two distinct types of tissue. The reflected wave travels off at an angle. The transmitted wave undergoes refraction only if two conditions are met: 1. The velocities of sound in the media forming the interface differ and 2. The incident angle is not perpendicular. 9 Figure 1.7 Comparison of specular and diffuse reflections. For diffuse reflection, echoes travel in various directions away from the interface so that an echo is not as dependent on interface orientation as for a specular reflector. In the human body, specular reflection is accompanied by diffuse reflection. 11 x Figure 1.8 Liver and kidney ultrasound scan. The boundaries of the organs are clearly defined as a result of specular and diffuse reflections happening at such interfaces. The liver appears as a hyperechoic (brighter) region due to its higher scattering level, whereas the kidney appears as a hypoechoic (darker) region due to its lower scattering level. (Courtesy of Philips Healthcare, Inc.) 12 Figure 1.9 Ocular globe wall scanned with a 10-MHz transducer (A) and a 20- MHz transducer (B). The episcleral space is better defined when imaged with the 20-MHz transducer (arrows). (Hewick et al., 2004). Although better spatial resolution is achieved with the 20-MHz transducer, better penetration is found at 10 MHz because of the higher attenuation at 20 MHz. 15 Figure 1.10 Direct piezoelectric effect: The piezoelectric material generates an electric potential when subjected to mechanical stress. (A) Under compression the piezoelectric material decreases its volume and the generated voltage has the same polarity as the piezoelectric material. (B) Under tension the piezoelectric material increases its volume and the generated voltage has opposite polarity as the piezoelectric material. Inverse piezoelectric effect: The piezoelectric material expands and contracts (i.e. vibrates) upon the application of an electric potential. (C) If the applied voltage has the same polarity as the piezoelectric material, this one expands. (D) If the applied voltage has the opposite polarity as the piezoelectric material, this one contracts. 16 Figure 1.11 The geometry of a piezoelectric material has an effect on its piezoelectric properties. This is illustrated by way of the electromechanical coupling coefficient: (A) a tall element of square cross-section frequently seen in composite materials, (B) a tall element frequently seen in linear arrays, and (C) a circular disc frequently seen in single-element transducers. (Shung, 2006) 17 Figure 1.12 Two different configurations of composite piezoelectric materials: (A) 1-3 composite, one phase of the composite is connected only in one direction whereas the second phase is connected in all three directions. (B) 2-2 composite, both phases are connected in two directions. (Newnham et al., 1978) 18 Figure 1.13 Cross-section views showing the detailed construction of two single- element ultrasonic transducers with one or two matching layers and backing. The transducer on the left has a lens, whereas the one on the right is self-focused. (Cannata et al., 2003) 19 xi Figure 1.14 Magnitude and phase of the input electrical impedance of a circular disc transducer irradiating into air and backed by air. (Shung, 2006) 20 Figure 1.15 Simulated pulse-echo impulse response and frequency spectrum of a 30-MHz single element transducer obtained via KLM modeling software. (PiezoCAD, Sonic Concepts, Woodinville, WA) 23 Figure 1.16 Axial resolution is the minimum distance between two reflectors positioned along the axis of an ultrasound beam that allows both reflectors to be resolved as separate objects. 25 Figure 1.17 Lateral resolution refers to how closely positioned two reflectors can be perpendicular to the beam axis and be seen as separate objects on an ultrasound image. 26 Figure 1.18 Sound pressure field as a function of distance (z) for a circular piston transducer under continuous wave (CW) excitation. (Courtesy of Olympus, Inc.) 27 Figure 1.19 Array transducers commercially available for clinical applications: (a) Linear sequential, (b) curvilinear sequential, (c) linear phased, (d) 1.5D phased and (e) 2D phased arrays. (Prince & Links, 2014) 30 Figure 1.20 Coordinate system depicting the difference in path length between the center element of a linear phased array and the nth element. (Shung, 2006) 32 Figure 1.21 The echoes returned from focal point P can be made to arrive simultaneously by applying delays to the echoes detected at the elements of the phased array. (Courtesy of Analog Devices, Inc.) 33 Figure 1.22 Discrete zone dynamic focusing is used for transmission, whereas almost continuous dynamic focusing may be used for reception in phased arrays. (Shung, 2006) 33 Figure 1.23 Dynamic aperture apodization (shading of the array elements) can be implemented to maintain the ultrasound beam width narrow and constant at all imaging depths. (Shung, 2006) 34 Figure 1.24 The amplitude of the input voltage signal can be uniform (left) or varied (right) across a transducer array aperture. An amplitude apodized signal is preferred over its uniform counterpart to help reduce the appearance of spurious side lobes. 35 xii Figure 1.25 Distinction between grating and side lobes. Grating lobes are indeed high side lobes that occur at certain angles due to constructive interference. In this case grating lobes are produced at ±40 by a linear array with an undesirable pitch. (Shung, 2006) 36 Figure 1.26 Comparison of beam profiles without (left) and with (right) spurious lobes. The beam at left is generated by six elements at 0.4 mm pitch and the beam at right by three elements at 1.0 mm pitch. (Courtesy of Olympus, Inc.) 36 Figure 1.27 Two false point targets are produced when steering a linear phased array transducer with spurious lobes at ±30 off the main lobe. The third picture on the right shows an example with a phantom composed of multiple wire point targets. (Lockwood et al., 1996) 37 Figure 1.28 Design parameters of linear sequential and linear phased arrays. 38 Figure 1.29 An ultrasound beam has three measurements of spatial resolution: Axial, lateral, and slice thickness resolution. (Penninck & d’Anjou, 2015) 39 Figure 1.30 Detailed construction isometric view of a linear array transducer where the plastic housing has been partially cut out to reveal the internal components. 40 Figure 1.31 Schematic depiction of the volume of tissue imaged during a B-mode scan done with a phased array. (Zagzebski, 1996) 41 Figure 2.1 Precancerous polyps (adenomas) grow within the lining of the colon and rectum. Over time, a few of these polyps can become cancerous (adenocarcinomas) and may penetrate the colon and rectum walls, spreading to lymph nodes along with other organs. (Courtesy of Blausen Medical, Inc.) 45 Figure 2.2 Cancer statistics for men and women in the United States. Colorectal cancer is the third most diagnosed cancer in both men and women, and the second leading cause of cancer related deaths of men and women together. (ACS and CDC, 2015) 45 Figure 2.3 Resection of colorectal tumor via TEM. (Courtesy of European School of Oncology) 46 Figure 2.4 The T stages of CRC. (Courtesy of Cancer Research United Kingdom) 47 xiii Figure 2.5 Colonoscopy is an endoscopic technique used to screen individuals for CRC. During a Colonoscopy procedure, a flexible fiber optic endoscope is introduced through the anus till the start of the colon. (Courtesy of Medical Billing and Coding Online) 48 Figure 2.6 Endorectal Ultrasound (ERUS) also known as Transrectal Ultrasound (TRUS). An ultrasound probe is inserted into the rectum for examination of the walls and tissue lying underneath. (Courtesy of WebMD, LLC) 49 Figure 2.7 Robotic surgery probe for real-time intraoperative fusion of ERUS and Endomicroscopy during TEM procedures. (Courtesy of the Hamlyn Centre for Robotic Surgery, Imperial College London) 51 Figure 2.8 CAD model of SL-ERUS linear phased array enclosed in protective 10F catheter. 52 Figure 2.9 CAD exploded view of array acoustic stack including backing, 2-2 array composite, 1 st matching layer (silver epoxy), and 2 nd matching layer (vapor deposited polymer) as well as zoomed view of separate layers and the final assembled array unit. The flexible circuit is included to show its positioning between the 2-2 array composite and the backing. 54 Figure 2.10 CAD schematics of flexible printed circuit design. 56 Figure 2.11 High-capacitance 48 AWG micro-coaxial cable. (Courtesy of Hitachi Metals, Ltd.) 57 Figure 2.12 2-Port Network representation of a micro-coaxial cable for estimating its scattering parameters: Reflection and Transmission Coefficients. 57 Figure 2.13 Reflection coefficient of the 48 AWG micro-coaxial cable at 1.5, 2.0, and 2.4 m. 58 Figure 2.14 Transmission coefficient of the 48 AWG micro-coaxial cable at 1.5, 2.0, and 2.4 m. 59 Figure 2.15 Verasonics Backshell Connector Kit. (Courtesy of Verasonics, Inc.) 61 Figure 2.16 Verasonics Vantage 128 System. 62 Figure 2.17 2D model of the SL-ERUS linear phased array with its x-y coordinates (symmetry boundary condition applied). 65 xiv Figure 2.18 Results obtained for representative piezoelectric element 33. Left: Voltage generated by received echo and its frequency spectrum. Right: Electrical impedance magnitude and electrical impedance phase angle. 70 Figure 2.19 Results obtained for crosstalk simulation: Voltage of transmitting element vs. first adjacent non-transmitting element (upper left), vs. second adjacent non-transmitting element (upper right), vs. third adjacent non-transmitting element (lower left), and vs. fourth adjacent non-transmitting element (lower right). 72 Figure 2.20 Computed crosstalk from 2 to 30 MHz. Maximum crosstalk levels at 15 MHz were: -20.8 dB for the first adjacent element, -23.9 dB for the second adjacent element, -24.75 dB for the third adjacent element, and -26.6 dB for the fourth adjacent element. 73 Figure 2.21 Time frame screenshot: Pressure wave being generated and propagating through the loading medium (top) and pressure history in the loading medium immediately in front of the center of the array transducer (bottom). 75 Figure 2.22 Pressure field in the loading medium of the SL-ERUS linear phased array under continuous wave excitation (100-cycle sinusoid with an amplitude of 100 Vpp at the design center frequency). Total loading medium length is 16 mm. 77 Figure 2.23 (A) Axial beam profile in the loading medium along the centerline of the array. (B) Lateral beam profile in the loading medium across the elevation natural focal plane at 8.1mm in front of the array. 77 Figure 2.24 Acoustic stack comprising the PMN-PT 2-2 composite and 1 st matching layer. Active and ground electrodes were sputtered on the bottom and top sides of the composite, respectively. Both layers were lapped down to final design thickness and the resulting stack was mechanically diced to final dimensions. 81 Figure 2.25 Flexible printed circuit for the SL-ERUS linear phased array. Left: Front side with cover-layer protected signal traces. Center: Back side with exposed common ground electrode featuring three windows for 2- 2 composite proper aligning and positioning. Right: Fitting test of flexible circuit folded into its final configuration and placed inside of a 10F polyimide tube. 82 xv Figure 2.26 Bonding the array acoustic stack to the flexible printed circuit. Left: Array acoustic stack placed on top of the flexible circuit. Center: Piezoelectric elements correctly aligned with the signal traces of the flexible circuit. Right: Pressure applied with a “C” clamp to bond the resulting sub-assembly. 83 Figure 2.27 (A) CAD rendered image of the array acoustic stack and backing layer bonded to the flexible circuit. (B) Bending the two grounding flaps in order to connect the ground layer on the backside of the flexible circuit to the front matching layer through metallized thru-vias. (C) Grounding flaps bonded to the front matching layer with conductive E-Solder 3022 epoxy. (D) Resulting sub-assembly with two main branches of the flexible circuit bonded to the sides of the backing layer with non- conductive Epo-Tek 301 epoxy. 84 Figure 2.28 (A) thru (C) Cross-sectional images revealing the diverse components of the array acoustic stack and backing layer bonded to the flexible circuit. (D) Detailed view of the bending of one main branch of the flexible circuit, making sure there were no broken signal traces. 85 Figure 2.29 Manual micro-soldering of sixty-four 48 AWG micro-coaxial cables to flexible printed circuit. (A) First trials showed broken coaxial cores and electrical shorts between consecutive signals. (B) 180-μm pitch soldering by hand is possible with no broken coaxial cores and 59 out of 64 soldered cores without electrical shorts. (C) Flexible circuit successfully fitted inside of a 10F catheter with its soldering branches folded into final configuration and with all the sixty-four 48 AWG micro-coaxial cables already soldered. 87 Figure 2.30 Cable assembly and packaging process. (A) All sixty-four 1.5-m long pieces of 48 AWG micro-coaxial cable carefully aligned and glued together inside of a 1-cm long piece of 8F PTFE tubing to which a 1F guide wire was hooked into. (B) All 64 pieces of micro-coaxial cable pulled through a 1.45-m long piece of 10F PTFE tubing. (C) Piece of 8F PTFE tubing cut off and other end of each micro-coaxial cable numbered. 88 Figure 2.31 Stripping specifications of 48 AWG micro-coaxial cable. 89 Figure 2.32 Connection of the 48 AWG micro-coaxial cable assembly to the flexible printed circuit. (A) Cores aligned and cold-soldered with E- Solder 3022 to the connection pads. All created connections were potted and protected with Epo-Tek 301. (B) Exposed outer conductors cold-soldered with E-Solder 3022 to the ground pads. (C) All 64 connections completed. 90 xvi Figure 2.33 Final packaging of the SL-ERUS phased array probe. (A) Soldering branches of the flexible printed circuit folded into final configuration. (B) Assembly pulled inside of the 10F PTFE catheter. (C) Finished distal end with 7-μm layer of vapor deposited Parylene C to ensure complete insulation. (D) Size comparison of the distal end of the SL- ERUS phased array probe, the array acoustic stack–backing layer– flexible circuit sub-assembly, and a dime coin. 91 Figure 2.34 Cable assembly termination of the proximal end. (A) 48 AWG micro- coaxial cables cold-soldered to Verasonics termination boards. (B) Two bundles of 32 micro-coaxial cables each held together with a piece of Kapton tape. (C) Protection with electrical shielding tape and black PVC cable-strain relief boot. (D) Finish with a Verasonics M14- A2-70 Hex cable-strain relief nut. 92 Figure 2.35 Connecting the SL-ERUS phased array probe to the Verasonics Backshell Connector. (A) Probe finished with Verasonics micro-coax termination boards. (B) Connection schematics of the termination boards to the left and right canister printed circuit boards. (C) Fitting Verasonics connection electronics inside one backshell half. (D) Completely finished and packaged SL-ERUS phased array probe. 93 Figure 2.36 (A) Simulated and (B) measured electrical impedance magnitude and phase angle of representative array element 33. 97 Figure 2.37 Uniformity of electrical impedance magnitude and phase angle values for all the array elements. 98 Figure 2.38 (A) Simulated and (B) measured pulse-echo response of representative array element 37. 100 Figure 2.39 Uniformity of -6 dB center frequency, -6 dB fractional bandwidth, and peak-to-peak sensitivities of all the array elements. 102 Figure 2.40 (A) Simulated and (B) measured crosstalk values for four nearest neighboring elements in the array. 103 Figure 2.41 Simulated and measured one-way azimuthal directivity response of representative array element 37. 104 Figure 2.42 Polished quartz reflector for imaging experiments of the SL-ERUS phased array probe. The polished quartz reflector was immersed in a tank with deionized water. 106 xvii Figure 2.43 Imaging of polished quartz reflector. (A) Target located in the near field at half the elevation natural focal distance. (B) Target located at the elevation natural focal distance. (C) Target located in the far field at twice the elevation natural focal distance. 107 Figure 2.44 Custom-made fine-wire phantom for imaging experiments of the SL- ERUS phased array probe. The phantom was immersed in a tank with deionized water. 108 Figure 2.45 Imaging of custom-made fine-wire phantom. (A) Acquired image. (B) Plot of the axial line spread function for the middle wire. (C) Plot of the lateral line spread function for the middle wire. 109 Figure 2.46 Custom-made tissue-mimicking phantom with inserted 700-μm- diameter stainless steel needle for imaging experiments of the SL- ERUS phased array probe. The phantom was immersed in a tank with deionized water. 110 Figure 2.47 Imaging of custom-made tissue-mimicking phantom with inserted stainless steel needle. (A) Needle inserted at a depth of 3 mm and oblique across the array transducer azimuth aperture. (B) Needle inserted at a depth of 5 mm and oblique across the array transducer azimuth aperture. (C) Needle inserted at a depth of 8 mm and oblique across the array transducer azimuth aperture. (D) Needle inserted at a depth of 8 mm and parallel across the array transducer elevation aperture. 111 Figure 2.48 Custom-made tissue-mimicking phantom with inserted piece of 3-mm- diameter polyimide tubing for imaging experiments of the SL-ERUS phased array probe. The phantom was immersed in a tank with deionized water. 112 Figure 2.49 Imaging of custom-made tissue-mimicking phantom with inserted piece of polyimide tubing. (A) Piece of polyimide tubing inserted at a depth of 5 mm and parallel across the array transducer elevation aperture. (B) Piece of polyimide tubing inserted at a depth of 8 mm and parallel across the array transducer elevation aperture. 113 Figure 3.1 Peripheral arterial disease (PAD) affects the vessels that carry blood from the heart to the limbs, being more common in the legs than in the arms. (Courtesy of the American Heart Association) 115 xviii Figure 3.2 Percutaneous access created into the femoral artery with a needle. The guide catheter-guide wire are then navigated through the vasculature to the origin of the blockage, to cross it afterwards, and set the guide wire in position for treatment delivery. (Adapted from the American Medical Center – American Heart Institute) 115 Figure 3.3 (A) Getting a guide catheter – guide wire across the blockage. (B) Inflating a balloon catheter to open the blocked artery, and (C) Inflating a balloon-stent catheter to deploy the metal stent in cases where the balloon angioplasty still leaves the artery partially blocked. (Adapted from the American Medical Center – American Heart Institute) 116 Figure 3.4 Angiogram showing blood flow in the superficial femoral artery (SFA). The blockage location is clearly visible; however, no images from inside the vessel can be obtained using this technology. Blindly navigating a catheter through the blockage with no intraluminal information (i.e. images) becomes a challenging task for vascular surgeons. (Adapted from WebMD) 117 Figure 3.5 State-of-the-art SL-IVUS catheters: (A) Rotating single element catheter, (B) Circular array catheter. (SL-IVUS blind spot image adapted from Volcano Corporation) 118 Figure 3.6 Incidence of adverse events in peripheral vascular interventions performed in the United States. 119 Figure 3.7 Effect of adventitial cuts on restenosis frequency for peripheral interventions via sub-intimal angioplasty. In a study involving 102 patients, 54% of them presented adventitial cuts with a concomitant restenosis frequency of 97%. (Krishnan et al., 2012) 120 Figure 3.8 FL-IVUS phased array catheter is capable of imaging stenosed areas in front of the catheter with no blind spots. 121 Figure 3.9 CAD model of FL-IVUS linear phased array enclosed in protective 8F catheter. 122 Figure 3.10 CAD schematics of flexible circuit design. 126 Figure 3.11 High-capacitance 48 AWG micro-coaxial cable. (Courtesy of Hitachi Metals, Ltd.) 126 Figure 3.12 2-Port Network representation of a micro-coaxial cable for estimating its scattering parameters: Reflection and Transmission Coefficients. 127 xix Figure 3.13 Reflection coefficient of the 48 AWG micro-coaxial cable at 0.6, 1.0, and 1.5 m. 128 Figure 3.14 Transmission coefficient of the 48 AWG micro-coaxial cable at 0.6, 1.0, and 1.5 m. 129 Figure 3.15 Verasonics Backshell Connector Kit. (Courtesy of Verasonics, Inc.) 130 Figure 3.16 Verasonics Vantage 128 System. 131 Figure 3.17 2D model of the FL-IVUS phased array catheter with its x-y coordinates (symmetry boundary condition applied). 133 Figure 3.18 Results obtained for representative piezoelectric element 17. Left: Voltage generated by received echo and its frequency spectrum. Right: Electrical impedance magnitude and electrical impedance phase. 134 Figure 3.19 Computed crosstalk from 10 to 50 MHz. Maximum crosstalk levels at 30 MHz were: -18.5 dB for the first adjacent element, -24.5 dB for the second adjacent element, -24.3 dB for the third adjacent element, and - 26.1 dB for the fourth adjacent element. 135 Figure 3.20 Pressure field in the loading medium of the FL-IVUS phased array catheter under continuous wave excitation (100-cycle sinusoid with an amplitude of 100 Vpp at the design center frequency). Total loading medium length is 10 mm. 136 Figure 3.21 (A) Axial beam profile in the loading medium along the centerline of the array. (B) Lateral beam profile in the loading medium across the elevation natural focal plane at 5 mm in front of the array. 136 Figure 3.22 (A) DRIE nickel mask for all 49 arrays with 2-2 composite pattern nested within a 1-3 composite pattern to prevent crystal fracture propagation in the sample. (B) Single array with 32-element aperture size of 0.8 mm x 1 mm. (C) 3 distinct individual sub-element lengths for each array ceramic pillar to prevent electrode strain that could result in open channel circuits. 141 Figure 3.23 Micrographs of three individual arrays as received from CTS Corporation. 142 xx Figure 3.24 Acoustic stack comprising the PMN-PT 2-2 composite and 1st matching layer. Active and ground electrodes were sputtered on the bottom and top sides of the composite, respectively. Both layers were lapped down to final design thicknesses and the resulting stack was mechanically diced to final dimensions. 143 Figure 3.25 Flexible printed circuit for the FL-IVUS phased array catheter. 144 Figure 3.26 Bonding the array acoustic stack to the flexible circuit. Left: Array acoustic stack placed on top of the flexible circuit. Center: Piezoelectric elements correctly aligned with the signal traces of the flexible circuit. Right: Pressure applied with a “C” clamp to bond the resulting sub-assembly. 145 Figure 3.27 (A) Array acoustic stack and backing layer bonded to the flexible circuit. (B) Bending the two branches of the flexible circuit and bonding them to the sides of the backing layer with non-conductive Epo-Tek 301 epoxy. (C) Resulting sub-assembly featuring increased contact area with the common ground electrode on the backside of the flexible circuit via conductive E-Solder 3022 epoxy. (D) Resulting sub- assembly surrounded by non-conductive Epo-Tek 301 epoxy and enclosed within a 2.5-mm long piece of 8F polyimide tubing. 147 Figure 3.28 (A) Cross-sectional image revealing the diverse components of the array acoustic stack and backing layer bonded to the flexible circuit. (B) Detail view of the bending of one branch of the flexible circuit, making sure there were no broken signal traces. 148 Figure 3.29 Cable assembly and packaging process. (A) All thirty-two 0.6-m long pieces of 48 AWG micro-coaxial cable carefully aligned and glued together inside of a 1-cm long piece of 6F PTFE tubing to which a 1F guide wire was hooked into. (B) All 32 pieces of micro-coaxial cable pulled through a 0.55-m long piece of 8F PTFE tubing, piece of 6F PTFE tubing cut off, and other end of each micro-coaxial cable numbered. 149 Figure 3.30 Stripping specifications of 48 AWG micro-coaxial cable. 150 xxi Figure 3.31 Connection of the 48 AWG micro-coaxial cable assembly to the flexible printed circuit. (A) Cores aligned and cold-soldered with E- Solder 3022 epoxy to their respective signal traces. All exposed outer conductors were also cold-soldered with each other using E-Solder 3022 epoxy. (B) All created connections were potted and protected with Epo-Tek 301 epoxy. (C) E-Solder 3022 epoxy applied to the backside of both branches of the flexible circuit as well as to the backside of the backing layer in order to create a continuous ground electrode. (D) E-Solder 3022 epoxy ground flaps created in order to strengthen the ground electrode connection on the front of the FL- IVUS probe. 151 Figure 3.32 Final packaging of the FL-IVUS phased array probe. (A) Finished distal end with 15-μm layer of vapor deposited Parylene C to serve as the 2 nd matching layer and to ensure complete insulation. (B) Size comparison of the finished distal end of the FL-IVUS phased array probe, the array acoustic stack–backing layer–flexible circuit sub- assembly, and a dime coin. 152 Figure 3.33 Cable assembly termination of the proximal end. (A) 48 AWG micro- coaxial cables cold-soldered to a Verasonics micro-coax termination board. (B) Protection with a black PVC cable-strain relief boot and finish with a Verasonics M14-A2-70 Hex cable-strain relief nut. 153 Figure 3.34 Connecting the FL-IVUS phased array probe to the Verasonics Backshell Connector. (A) Probe finished with Verasonics micro-coax termination board. (B) Connection schematics of the termination board to the left canister printed circuit board. (C) Fitting Verasonics connection electronics inside one backshell half. (D) Completely finished and packaged FL-IVUS phased array probe. 154 Figure 3.35 Flexible printed circuit for electrical impedance measurements. (A) CAD schematics. (B) Physical sample. (C) Array acoustic stack and 25 AWG ground wire bonded to flexible printed circuit. 156 Figure 3.36 (A) Simulated and (B) measured electrical impedance magnitude and phase angle of representative array element 17. 157 Figure 3.37 Uniformity of electrical impedance magnitude and phase angle values for all the array elements. 158 Figure 3.38 (A) Simulated and (B) measured pulse-echo response of representative array element 22. 160 Figure 3.39 Uniformity of -6 dB center frequency, -6 dB fractional bandwidth, and peak-to-peak sensitivity of all the array elements. 162 xxii Figure 3.40 (A) Simulated and (B) measured crosstalk values for four nearest neighboring elements in the array. 163 Figure 3.41 Simulated and measured one-way azimuthal directivity response of representative array element 22. 165 Figure 3.42 Polished quartz reflector for imaging experiments of the FL-IVUS phased array probe. The polished quartz reflector was immersed in a tank with deionized water. 167 Figure 3.43 Imaging of polished quartz reflector. (A) Target located in the near field at half the elevation natural focal distance. (B) Target located at the elevation natural focal distance. (C) Target located in the far field at twice the elevation natural focal distance. 168 Figure 3.44 Custom-made fine-wire phantom for imaging experiments of the FL- IVUS phased array probe. The phantom was immersed in a tank with deionized water. 169 Figure 3.45 Imaging of custom-made fine-wire phantom. (A) Acquired image. (B) Plot of the axial line spread function for the middle wire. (C) Plot of the lateral line spread function for the middle wire. 170 Figure 3.46 Custom-made porcine carotid artery phantom for imaging experiments of the FL-IVUS phased array probe. The phantom was immersed in a tank with deionized water. 171 Figure 3.47 Imaging of custom-made porcine carotid artery phantom – widest section, 3-mm diameter. (A) Occlusion-free arterial lumen. (B) Half- occluded arterial lumen. (C) Increased level of occlusion, only a narrow free passage on the bottom right section of the arterial lumen. (D) Completely occluded arterial lumen. 172 Figure 3.48 Imaging of custom-made porcine carotid artery phantom – transition section, from 3-mm to 2.5-mm diameter. (A) Occlusion-free arterial lumen. (B) Occlusion leaving only a narrow free passage on the bottom left section of the arterial lumen. 173 Figure 3.49 Imaging of custom-made porcine carotid artery phantom – transition section, from 2.5-mm to 1-mm diameter. (A) Occlusion-free arterial lumen. (B) Occlusion leaving only a narrow free passage on the bottom left section of the arterial lumen. 173 xxiii LIST OF TABLES Table I Change in ultrasound beam direction for a 30° angle of incidence. 10 Table II Ultrasound attenuation coefficients of various tissues. 14 Table III Main characteristics of array transducers. (Shung, 2006; Prince & Links, 2014) 31 Table IV Design rules of linear sequential and linear phased arrays. (Shung, 2006) 38 Table V SL-ERUS linear phased array design parameters. 54 Table VI Material properties of the SL-ERUS linear phased array acoustic stack. 55 Table VII Verasonics Vantage 128 System technical specifications. 62 Table VIII SL-ERUS linear phased array design parameters and simulated performance. 78 Table IX Comparison of simulated and measured electrical impedance results. 97 Table X Comparison of simulated and measured pulse-echo response results. 101 Table XI Comparison summary of simulated and measured characterization results. 105 Table XII FL-IVUS linear phased array design parameters. 124 Table XIII Material properties of the FL-IVUS linear phased array acoustic stack. 124 Table XIV Verasonics Vantage 128 System technical specifications. 132 Table XV FL-IVUS phased array catheter design parameters and simulated performance. 137 Table XVI Comparison of simulated and measured electrical impedance results. 158 Table XVII Comparison of simulated and measured pulse-echo response results. 162 Table XVIII Comparison summary of simulated and measured characterization results. 166 xxiv Table XIX SL-ERUS linear phased array design parameters and comparison summary of simulated and measured performance results. 175 Table XX FL-IVUS linear phased array design parameters and comparison summary of simulated and measured performance results. 178 1 CHAPTER 1 INTRODUCTION This chapter introduces ultrasound as a medical imaging modality and discusses its advantages over other medical imaging alternatives. A brief overview of the latest developments in medical ultrasound technology together with their safety requirements is presented to familiarize the reader with the diversified scope of applications of medical ultrasound imaging. Basic ultrasound phenomenon principles are then discussed together with design considerations and performance characterization of single-element and array ultrasound transducers. The later sections of this chapter outline current trends in medical ultrasound imaging using array transducers and describe the objectives of this research. 1.1 Medical Ultrasound Medical ultrasound is an imaging modality with wide applicability that does not only complement the more traditional approaches like X-rays, but also possesses unique characteristics that confers it advantages over other competing modalities such as X-ray computed tomography (CT), positron emission tomography (PET), single-photon emission computed tomography (SPECT), and magnetic resonance imaging (MRI). Namely, such advantages include its real-time imaging capability, portability, and safeness. Medical ultrasound is the only imaging modality capable of producing images in real time and requires less space and capital expenditure than other medical imaging modalities of similar capabilities (Shung, 2006). Real-time imaging gives ultrasound users much more comfort 2 in the diagnosis of moving targets, whether it is a slow object like a fetus in the third trimester or a fast one like the valves of the heart (Kim, 2010). From the patients’ perspective this is another convenient advantage as they do not have to be kept immobilized for prolonged periods of time. In addition to visual information, medical ultrasound can also provide flow and audible information. For instance, in diagnosing mitral valve regurgitation, as shown in Figure 1.1, the mitral valve was imaged with 2D grayscale mode, the blood flow was detected by color flow Doppler mode, and the direction and speed of the blood flow was quantized by the pulsed Doppler mode. Additionally, the audio sound of the blood flow is given to the clinician by speakers attached to the system. This multi-mode, combined information enables a more accurate diagnosis of the medical condition and therefore leads to better clinical outcomes. Figure 1.1. 2D/Color/Pulsed Wave Doppler combined mode image of the heart showing mitral valve regurgitation. (Courtesy of Peter Munk Cardiac Centre) 3 Medical ultrasound is portable and can be easily transported to the bedside of a patient at a hospital (Figure 1.2.A). Recently, such portability reached the handheld device realm with the introduction of the Vscan TM ultrasound system by General Electric Healthcare in early 2010. The unit weighs one pound and is only 3 inches wide and 5.3 inches long, offering real-time, 2D, black and white imaging and color-coded flow Doppler mode on a 240 x 320 pixel LCD screen. The latest version (Figure 1.2.B) operates for up to 90 minutes on a fully charged battery and offers a dual probe featuring a low-frequency phased array transducer on one end of the probe for deep targets (e.g. heart and abdomen) and a high-frequency linear array transducer on the other end of the probe for shallow targets (e.g. lungs, arteries, and long bones). Its enhanced portability allows the Vscan TM to easily be transported to locations where having a device continuously plugged in to the power grid may not be possible, as in the case of rural clinics in third world and developing countries (Figure 1.2.C). (A) (B) (C) Figure 1.2 (A) GE Vivid S6 TM ultrasound system being used in a hospital setting. The GE Vivid S6 TM delivers imaging and analysis in a broad range of applications: Cardiac, vascular, abdominal, and pediatric/fetal (adapted from General Electric Healthcare, Inc.). (B) GE Vscan TM pocket-sized, handheld (dual probe version) ultrasound system and (C) GE Vscan TM (phased array single probe version) being used in a rural clinic. (Courtesy of General Electric Healthcare, Inc.) 4 In contrast to X-rays, CT, PET, SPECT, and MRI, medical ultrasound is the only imaging modality that does not produce any ionizing radiation, making it the most biologically safe medical imaging modality currently available (Shung, 2006). Moreover, to ensure ultrasound safeness, its power level and intensity limits are closely regulated by authorities such as the International Electrotechnical Commission (IEC) and the Food and Drug Administration (FDA) (Szabo, 2013). Additionally, compliant to the American Institute of Ultrasound in Medicine (AIUM) and the National Electrical Manufacturers Association (NEMA) Output Display Standard (ODS), medical ultrasound systems with Doppler mode capability must display the thermal (TI) and mechanical (MI) indices in real time (Marsal, 2005). These indices give medical practitioners an estimation of the temperature increase in the tissue and the likelihood of cavitation by ultrasound, respectively. Figure 1.3 below shows a color Doppler image of the umbilical cord with the TI and MI indices displayed at the upper right corner. Based on the image and the type of TI used (thermal index for bone – TIb), it can be inferred this is a second or third trimester fetus and the scanning time is less than 30 minutes (0.5 < TIb < 1). Since the umbilical cord is being imaged (i.e. no gas bodies present) the MI can be > 0.4 but kept as low as possible in order to comply with the ALARA (As Low As Reasonably Achievable) principle for patient scanning (Nelson et al., 2009). 5 Figure 1.3. 2D color Doppler mode image of the umbilical cord showing real-time display of thermal (TI) and mechanical (MI) indices at the upper right corner. (Courtesy of Philips Healthcare, Inc.) Overall, real-time imaging capability, portability, and safeness, among many other advantages, enable medical ultrasound to be ranked as the 2 nd diagnostic imaging modality in terms of the number of clinical cases that use it, just below conventional X-ray imaging (Shung, 2006). 1.2 Ultrasound Principles Ultrasound is sound generated above the human hearing range, typically considered from 20 to 20,000 Hz. However, the frequency range normally employed in conventional low- frequency medical ultrasound is 2 to 15 MHz, whereas that of high-frequency medical ultrasound 6 is generally 15 to 120 MHz (Kim, 2010), with some reported applications going beyond 150 MHz (Yoon et al., 2015). Although ultrasound behaves similarly to audible sound, it has a much shorter wavelength, which means that it can be reflected off very small surfaces such as irregularities and defects present in any type of materials (e.g. bones and tissues). It is this property, together with the aforementioned advantages in Section 1.1, which makes ultrasound useful for medical imaging applications. Ultrasound vibrations travel in the form of a wave only through an elastic medium such as a liquid or a solid (i.e. no travel through vacuum). Shown in Figure 1.4 are the basic parameters of a continuous wave (CW). These parameters include the wavelength (λ) at a fixed time and the period (T) at a fixed distance of a complete cycle. The back-and-forth displacement of the ultrasound source squeezes and pulls on the particles in the medium, which in turn push and pull on neighboring particles. Regions where the particles are pushed together are called regions of compression, whereas regions where the particles are drawn apart are called regions of rarefaction. Figure 1.4. CW excitation of an elastic medium by an ultrasound source. 7 The velocity of ultrasound (c), measured in m/sec, in a perfectly elastic material at a given temperature and pressure is constant. The relation between c, λ, and T is given by equation (1.1), which can be likewise expressed as in equation (1.2), where the frequency term (f), measured in Hertz (Hz), is the mathematical reciprocal of T and corresponds to the number of cycles completed in one second. 𝑐 = λ 𝑇 (1.1) 𝑐 = 𝜆𝑓 (1.2) The most common methods of ultrasound imaging utilize either longitudinal waves or shear waves (Figure 1.5). Other forms of ultrasound propagation exist, including surface waves and Lamb waves. The longitudinal wave is a compressional wave in which the particle motion is in the same direction as the propagation of the wave. Ultrasound waves that travel through tissue are longitudinal waves. The shear (transverse) wave presents particle motion perpendicular to the direction of the wave propagation. Shear waves can propagate easily through some solid materials (e.g. bones); however, they do not travel effectively through soft tissue. Consequently, longitudinal waves are the most important in diagnostic ultrasound imaging (Zagzebski, 1996). Surface (Rayleigh) waves present an elliptical particle motion and travel across the surface of a target. Their velocity is approximately 90% of the shear wave velocity of the target being studied and their penetration depth is approximately equal to one wavelength. Lastly, plate (Lamb) waves present a complex vibration occurring in materials where the thickness is less than one wavelength of the ultrasound applied onto it. 8 Figure 1.5. Particle motion versus the direction of wave propagation for longitudinal and shear waves. Longitudinal waves are the most commonly employed in medical ultrasound imaging. (Courtesy of Olympus, Inc.) An important parameter that influences the amplitude of reflected echoes as ultrasound waves travel through distinct layers of tissue is a quantity called acoustic impedance (Z). The acoustic impedance of a medium is the opposition to displacement of its particles by sound and is defined as: 𝑍 ±= 𝑝 ± 𝑢 𝑧 ± = ±𝜌𝑐 (1.3) where p is the ultrasound wave pressure, uz is the resultant particle velocity in the z-direction, ρ is the medium’s density, and c is the medium’s velocity of sound as previously defined by equations (1.1) and (1.2). The product ρc is also called the characteristic impedance of the medium and is measured in kg/m 2 -sec, or rayl, in honor of Lord Rayleigh, the father of modern acoustics. Whenever an ultrasound wave is incident on an interface between two media having different acoustic impedances, some of the energy of the incident wave is reflected and the rest is 9 transmitted into the second medium. The amplitude of the reflected echo depends on the difference between the acoustic impedances of the two media forming the interface (e.g. two distinct types of tissue). The directions of the reflected and transmitted waves are governed, just as in optics, by Snell’s law. This is shown in Figure 1.6, where the subscripts i, r, and t refer to incident, reflected, and transmitted waves, respectively. Figure 1.6. Reflection and transmission of a plane ultrasonic wave incident on an interface between two distinct types of tissue. The reflected wave travels off at an angle. The transmitted wave undergoes refraction only if two conditions are met: 1. The velocities of sound in the media forming the interface differ and 2. The incident angle is not perpendicular. As in optics, ϴi = ϴr and (sin ϴi /sin ϴt) = c1/c2. For any incident angle greater than the critical angle (ϴic), where ϴic = arcsin(c1/c2), if c2 >c1 there is no transmission and total reflection occurs. If c2 <c1 total reflection does not occur even at very shallow incident angles. The pressure reflection (R) and transmission (T) coefficients are defined as: 𝑅 = 𝑝 𝑟 𝑝 𝑖 = 𝑍 2 cos θ 𝑖 −𝑍 1 cos θ 𝑡 𝑍 2 cos θ 𝑖 +𝑍 1 cos θ 𝑡 (1.4) 10 𝑇 = 𝑝 𝑡 𝑝 𝑖 = 2𝑍 2 cos θ 𝑡 𝑍 2 cos θ 𝑖 +𝑍 1 cos θ 𝑡 (1.5) For perpendicular incidence, ϴi = ϴt = 0 and equations (1.4) and (1.5) become: 𝑅 = 𝑝 𝑟 𝑝 𝑖 = 𝑍 2 −𝑍 1 𝑍 2 +𝑍 1 (1.6) 𝑇 = 𝑝 𝑡 𝑝 𝑖 = 2𝑍 2 𝑍 2 +𝑍 1 (1.7) Understanding how a transmitted ultrasound beam undergoes refraction while performing a patient’s scanning results of utmost importance to the trained eye of clinicians and ultrasonographers. Only those with enough experience can accurately diagnose the medical condition and consequently lead the patient to a better clinical outcome. Table I lists results for some soft tissue interfaces, of which the muscle-to-fat interface presents the severest refraction. Indeed, fat and adipose tissue often pose difficulty to ultrasound imaging; some of this difficulty is believed to be from refraction that happens at adipose tissue interfaces (Zagzebski, 1996). Table I. Change in ultrasound beam direction for a 30° angle of incidence. Interface Transmitted beam angle Angle change Muscle-fat 27.1° 2.9° Muscle-fluid 28.8° 1.2° Muscle-blood 29.2° 0.8° When speaking about reflection it is useful to distinguish between specular and diffuse reflections. Specular reflection refers to echo signals produced by large and smooth interfaces having dimensions much greater than the ultrasonic wavelength, as it is usually the case of a water-quartz crystal interface. In the human body, specular reflection alone does not happen as all tissue interfaces are not perfectly specular, meaning that they are not perfectly smooth and 11 possess a degree of roughness (e.g. walls of the kidneys and liver). For this reason, specular reflection in the human body is accompanied by diffuse reflection, which refers to echo signals produced by interfaces having dimensions on the order of the ultrasonic wavelength (Robinson et al., 1981). Indeed, diffuse reflection is beneficial for medical ultrasound as diffuse echo signals travel in various directions away from the interface so that an echo is not as dependent on interface orientation as for a specular reflector (Figure 1.7). Figure 1.7. Comparison of specular and diffuse reflections. For diffuse reflection, echoes travel in various directions away from the interface so that an echo is not as dependent on interface orientation as for a specular reflector. In the human body, specular reflection is accompanied by diffuse reflection. Ultrasonic scattering is another source of echo signals in the human body arising from small objects on the order of the ultrasonic wavelength or smaller (e.g. reflectors present in the parenchyma of most organs and red blood cells). Similarly to diffuse reflection, with ultrasonic scattering the echoes tend to travel off in all directions having little orientation dependence of the echo signal strength, with the difference that echoes from scattering tend to be weaker than those from specular and diffuse reflections happening at organ boundaries. In fact, ultrasonic scattering gives rise to much of the diagnostic information seen in medical ultrasound imaging. Changes in scattering echo amplitude from one region to another result in brightness changes on an ultrasound image; therefore they are useful in delineating different organs and even abnormal structures within an organ (Meuwly and Gudinchet, 2004). Regions that appear brighter are 12 called hyperechoic and result from increases in the ultrasound scattering level compared to the surrounding tissue; whereas hypoechoic regions appear darker as a result of a lower ultrasound scattering level compared to the surrounding tissue. Figure 1.8 shows a liver and kidney ultrasound scan where the boundaries of such organs are clearly defined due to specular and diffuse reflections occurring at such interfaces. The parenchyma of each organ has a different scattering level, being higher in the liver (hyperechoic) than in the kidney (hypoechoic). Figure 1.8. Liver and kidney ultrasound scan. The boundaries of the organs are clearly defined as a result of specular and diffuse reflections happening at such interfaces. The liver appears as a hyperechoic (brighter) region due to its higher scattering level, whereas the kidney appears as a hypoechoic (darker) region due to its lower scattering level. (Courtesy of Philips Healthcare, Inc.) As an ultrasound beam travels through layers of tissue in the human body, one part of its energy is lost and redistributed via the previously discussed processes of reflection and scattering, and another part is absorbed by the medium. The energy absorbed by the medium is also lost and converted to heat. The absorption mechanisms in biological tissues are quite complex and have been assumed to arise from classical absorption due to viscosity and a 13 relaxation phenomenon (Shung, 2006). Nevertheless, under ordinary circumstances with diagnostic ultrasound, the amount of heat produced is too small to cause a considerable temperature change (Zagzebski, 1996). Altogether, the aforementioned phenomena (1. reflection and scattering, and 2. absorption) decrease the amplitude and intensity of an ultrasound beam as a function of distance as it traverses tissue. The general term for this decrease in amplitude with increased distance traveled is attenuation. Equation (1.8) is utilized to find the pressure amplitude of a plane monochromatic wave propagating in the z-direction: 𝑝 (𝑧 ) = 𝑝 (𝑧 = 0)𝑒 −𝛼𝑧 (1.8) where p(z = 0) is the pressure amplitude at z = 0 and α is the pressure attenuation coefficient. Therefore, the attenuation coefficient is expressed as: 𝛼 = 1 𝑧 ln[ 𝑝 (𝑧 =0) 𝑝 (𝑧 ) ] (1.9) The attenuation coefficient has a unit of nepers per centimeter (np/cm) and is sometimes expressed in units of decibels per centimeter (dB/cm), where α(dB/cm) = 8.686α(np/cm). Attenuation in soft tissues is highly dependent on the ultrasonic frequency. In most cases, attenuation is nearly proportional to the frequency, thus if the attenuation coefficient for a tissue is given at a frequency of 1 MHz, it doubles at 2 MHz, triples at 3 MHz, and so on. On this basis, to determine the amount of attenuation occurring when an ultrasound beam passes through a given thickness of tissue, the following expression is utilized: Attenuation = 𝛼 × 𝑙 × 𝑛 (1.10) 14 where l is the traversed thickness of tissue and n is a scalar factor that results from dividing the frequency of interest by the frequency at which the attenuation coefficient is given. Typical values reported for the ultrasonic attenuation coefficients in soft tissues are shown in Table II (Goss et al., 1978). In general, it has been found that for most soft tissues the attenuation coefficient is near 0.5 to 1 dB/cm at 1 MHz (Ophir et al., 1982). Table II. Ultrasound attenuation coefficients of various tissues. Tissue Attenuation at 1 MHz (dB/cm) Water 0.0022 Blood 0.19 Liver 0.5 Muscle 1.15 Fat 0.48 Marrow 0.5 Breast 0.75 Figure 1.9 shows two images of the ocular globe wall and the episcleral space. The episcleral space separates the globe wall from the orbit and is seen as a thin dark line. It is an important ocular landmark often used to determine the sensitivity of B-scanners and their suitability for ophthalmic use. As it can be noticed, the episcleral space is better defined when imaged with a 20-MHz transducer (B) than with a 10-MHz transducer (A). (Hewick et al., 2004). 15 Figure 1.9. Ocular globe wall scanned with a 10-MHz transducer (A) and a 20-MHz transducer (B). The episcleral space is better defined when imaged with the 20-MHz transducer (arrows). (Hewick et al., 2004). Although better spatial resolution is achieved with the 20-MHz transducer, better penetration is found at 10 MHz because of the higher attenuation at 20 MHz. Spatial resolution in ultrasound is improved when imaging is done at high ultrasound frequencies. However, high ultrasound frequencies are accompanied by high attenuation losses, resulting in poorer penetration into tissue. When designing single-element and array ultrasound transducers, a compromise has to be made between attenuation losses in tissue and spatial resolution requirements according to the clinical application. 1.3 Ultrasound Transducers The general term transducer refers to any device that is used to convert signals or energy from one form to another. Transducers used in medical ultrasound employ the piezoelectric effect to generate sound waves and detect echo signals. The piezoelectric (pressure-electric) effect is a phenomenon in which a material, upon the application of an electrical field, changes its physical dimensions and vice versa (Cady, 1964; Kino, 1987). The piezoelectric effect was discovered in the 1880s by the French physicists Pierre and Jacques Curie; they found that when 16 a force is applied perpendicularly to the faces of a quartz crystal, an electrical charge results. The direct and inverse piezoelectric effects are illustrated in Figure 1.10. At present, naturally occurring piezoelectric crystals are rarely used as transducer materials in medical ultrasound imaging due to their weak piezoelectric properties. Two of the most popular materials are a polycrystalline ferroelectric ceramic, lead zirconate titanate (PZT), and a single crystal ferroelectric ceramic, lead magnesium niobate-lead titanate (PMN-PT), which possess excellent piezoelectric properties following polarization. Polarization of a piezoelectric material is done by heating it to a temperature just above its Curie temperature and letting it to cool slowly in the presence of a strong electric field, typically in the order of 20 kV/cm applied in the direction in which the piezoelectric effect is required (Shung, 2006). Figure 1.10. Direct piezoelectric effect: The piezoelectric material generates an electric potential when subjected to mechanical stress. (A) Under compression the piezoelectric material decreases its volume and the generated voltage has the same polarity as the piezoelectric material. (B) Under tension the piezoelectric material increases its volume and the generated voltage has opposite polarity as the piezoelectric material. Inverse piezoelectric effect: The piezoelectric material expands and contracts (i.e. vibrates) upon the application of an electric potential. (C) If the applied voltage has the same polarity as the piezoelectric material, this one expands. (D) If the applied voltage has the opposite polarity as the piezoelectric material, this one contracts. The piezoelectric properties of a material depend on its boundary conditions and shape. For example, the piezoelectric constant of a material in the plate form is different from that in the 17 rod or disc forms. Their ability to convert one form of energy into another is measured by its electromechanical coupling coefficient, defined as stored mechanical energy divided by total stored energy. Because only the stored mechanical energy is useful, the electromechanical coupling coefficient is a measure of the performance of a material as a transducer. Figure 1.11 shows three of the geometries that are frequently used in ultrasonic imaging with the effect of geometry on the electromechanical coupling coefficient. Figure 1.11. The geometry of a piezoelectric material has an effect on its piezoelectric properties. This is illustrated by way of the electromechanical coupling coefficient: (A) a tall element of square cross-section frequently seen in composite materials, (B) a tall element frequently seen in linear arrays, and (C) a circular disc frequently seen in single-element transducers. (Shung, 2006) Transducer elements do not necessarily consist in pure piezoelectric materials only; nowadays, piezoelectric composite materials are used in many transducers and ultrasound scan heads. These composites are mixtures of piezoelectric ceramics with other materials, such as non-conductive epoxy. A composite is formed by cutting grooves into the face of a piezoelectric sample, leaving behind piezoelectric posts, and then filling the space between the posts with non- conductive epoxy. Composites have several advantages over pure piezoelectric ceramics: They are lighter in weight, they have a lower acoustic impedance that makes it easier to match the 18 impedance of the transducer to that of tissue, transducers with piezoelectric composites have wider frequency bandwidths, and their higher coupling coefficient together with better impedance matching can lead to higher transducer sensitivity and improved image resolution. Composites are commonly available in two configurations: 1-3 composite and 2-2 composite (Figure 1.12). One of the problems associated with composite piezoelectric materials, however, is the higher fabrication cost and complexity. Figure 1.12. Two different configurations of composite piezoelectric materials: (A) 1-3 composite, one phase of the composite is connected only in one direction whereas the second phase is connected in all three directions. (B) 2-2 composite, both phases are connected in two directions. (Newnham et al., 1978) The simplest ultrasonic transducer for medical imaging applications is a single-element piston transducer as shown in Figure 1.13 (Lockwood et al., 1994; Cannata et al., 2003). In such transducer, the surfaces of the piezoelectric element are sputtered with a dual layer of chrome and gold and the outside electrode is grounded to the metallic housing to protect patients from electrical shock. Acoustic insulating epoxy is placed between the piezoelectric element and the housing to prevent ringing of the housing that follows the vibration of the piezoelectric element. 19 Figure 1.13. Cross-section views showing the detailed construction of two single-element ultrasonic transducers with one or two matching layers and backing. The transducer on the left has a lens, whereas the one on the right is self-focused. (Cannata et al., 2003) An ultrasonic transducer has a resonance frequency at which it is most efficient in converting electrical energy to acoustic energy and vice versa. The resonance frequency is determined mainly by the thickness of the piezoelectric element, being the relationship between these two parameters given by the following expression: 𝑓 𝑜 = 𝑛 𝑐 𝑝 2𝐿 (1.11) where cp is the acoustic wave velocity in the transducer material, L is the thickness of the piezoelectric material, and n is an odd integer with the lowest resonant frequency being at n = 1. In other words, resonance occurs when L is equal to odd multiples of one-half wavelength or: 𝐿 = 𝑛 𝜆 𝑝 2 (1.12) where λp is the wavelength in the piezoelectric material. The resonance frequency of an ultrasonic transducer can also be obtained from the plot of its input electrical impedance magnitude and phase (Figure 1.14). At resonance, or series frequency (fr or fs), the input 20 electrical impedance magnitude is minimal and the phase angle changes from -90° to 90°. At anti-resonance, or parallel frequency (fa or fp), the input electrical impedance magnitude is maximal and the phase angle changes from 90° to -90°. Additionally, by knowing the resonance and anti-resonance frequencies, the electromechanical coupling coefficient of the piezoelectric material, kt (numerical measure of the conversion efficiency between electrical and acoustic energy), can be calculated as follows (Krimholtz et al., 1970): 𝑘 𝑡 = √ 𝜋 2 𝑓 𝑟 𝑓 𝑎 tan( 𝜋 2 𝑓 𝑎 −𝑓 𝑟 𝑓 𝑎 ) (1.13) Figure 1.14. Magnitude and phase of the input electrical impedance of a circular disc transducer irradiating into air and backed by air. (Shung, 2006) 21 Ultrasound transducers are driven by short duration electrical pulses from a pulse transmitter. In response to these excitation pulses, the transducer piezoelectric element vibrates at its resonance frequency emitting a pulse of sound into the medium. For most applications it is desirable to emit ultrasound pulses of very short duration to optimize the axial resolution of the transducer. In order to meet this condition and avoid the so-called ringing effect for pulse-echo applications, absorptive backing materials with an acoustic impedance similar to that of the piezoelectric material can be used to damp out the ringing and increase the transducer bandwidth, at expense of sacrificing sensitivity as a large portion of the energy is absorbed by the backing material. Backing materials include, but are not limited to, tungsten-loaded epoxies, silver-loaded epoxies, and liquid plasticizer-loaded epoxies (Wang et al., 2001). In order to improve their partly lost sensitivity due the presence of a backing, most transducers have acoustic matching layers in the front. These matching layers provide efficient transmission of sound waves from the transducer piezoelectric element to soft tissue and vice versa. They do this by reducing reflections at the transducer-tissue interface. For a monochromatic plane wave, 100% transmission occurs for a layer of material having a quarter wavelength thickness and acoustic impedance determined by the expression: 𝑍 𝑚 = √ 𝑍 𝑝 𝑍 𝑙 (1.14) where Zp and Zl are the acoustic impedances of the piezoelectric element and the loading medium, respectively (Kinsler et al., 1982). For wideband transducers the previous expression for a single matching layer is improved to (Desilets et al., 1978): 𝑍 𝑚 = (𝑍 𝑝 𝑍 𝑙 2 ) 1/3 (1.15) 22 It is important to mention that a single matching layer is exactly one quarter wavelength only for one frequency. Thus, all frequencies in a pulsed waveform are not efficiently transmitted with a single matching layer. Having multiple matching layers provides efficient ultrasound transmission between the piezoelectric element and soft tissue for a range, or spectrum, of ultrasound frequencies. For two matching layers, the acoustic impedances of the two layers are respectively determined by (Desilets et al., 1978): 𝑍 𝑚 1 = (𝑍 𝑝 4 𝑍 𝑙 3 ) 1/7 (1.16) 𝑍 𝑚 2 = (𝑍 𝑝 𝑍 𝑙 6 ) 1/7 (1.17) Well-designed matching layers can ease the requirements previously mentioned for the backing layer. Unlike the backing layer, which must absorb the energy removed from the piezoelectric element, therefore wasting it, the matching layers provide very efficient use of the ultrasonic energy. The energy transmitted out the front of the transducer contributes to the useful ultrasound beam. Hence, the best design approach consists in using an optimized acoustic matching in the front and a light backing layer in the back. This combination results in excellent performance in terms of both bandwidth and sensitivity (Desilets et al., 1978). The bandwidth of an ultrasound transducer may be determined by spectral analysis of its ultrasound pulse. Figure 1.15 shows the ultrasound pulse and spectral plot of a 30-MHz single element transducer simulated via KLM modeling software (PiezoCAD, Sonic Concepts, Woodinville, WA). 23 Figure 1.15. Simulated pulse-echo impulse response and frequency spectrum of a 30-MHz single element transducer obtained via KLM modeling software. (PiezoCAD, Sonic Concepts, Woodinville, WA) The spectral plot represents the fraction of signal within a given frequency interval versus the resonance center frequency, which usually corresponds to the frequency at which the spectrum has its maximum. The transducer bandwidth is commonly determined using the frequencies at which the fraction of signal equals to ½ its original value, which is the same as a change of -6 dB. The transducer sensitivity is determined with the arithmetic addition of the absolute peak voltage values of the pulse, and the pulse width or pulse duration is commonly determined using the length of time at which the fraction of signal equals to either ½ or 1 /10 its maximum absolute peak, which is a change of either -6 or -20 dB. Another measure of the frequency bandwidth is the transducer quality factor, or Q-factor. It is defined as: 𝑄 = 𝑓 0 (𝑓 2 −𝑓 1 ) (1.18) 24 where f0 is the transducer resonance center frequency and f2 and f1 are the frequencies defining the bandwidth of the transducer. For a fixed center frequency, the narrower the frequency bandwidth, the higher the Q. High Q transducers are used in applications in which maximal output is needed and bandwidth requirement is not crucial, as in continuous-wave Doppler echocardiography for the noninvasive assessment of left ventricular outflow tract pressure gradient (Panza et al., 1992). However, the Q should be low in transducers used in pulse-echo ultrasound imaging devices, as in angled-focused single element transducers for intravascular ultrasound imaging (Yoon et al., 2015). As previously stated in section 1.2 Ultrasound Principles, spatial resolution in ultrasound is improved when imaging is done at high ultrasound frequencies. However, high ultrasound frequencies are accompanied by high attenuation losses, resulting in poorer penetration into tissue. The choice of ultrasound frequency for any medical examination is the result of a compromise between resolution requirements and the ability to obtain satisfactory sound beam penetration to image all the tissues of interest. In ultrasound, spatial resolution is determined via two main parameters: Axial resolution and lateral resolution. Axial resolution refers to the minimum reflector spacing along the axis of an ultrasound beam that results in separate, distinguishable echoes on a display as shown below in Figure 1.16. 25 Figure 1.16. Axial resolution is the minimum distance between two reflectors positioned along the axis of an ultrasound beam that allows both reflectors to be resolved as separate objects. Axial resolution is most closely related to the pulse duration, which can be commonly determined using the length of time at which the fraction of signal equals to a change of either -6 or -20 dB of the maximum absolute peak, as aforementioned. Alternatively, pulse duration (PD) can also be determined using the following expression: 𝑃𝐷 = 𝑁 𝑐 𝑓 (1.19) where Nc is the number of cycles in the pulse and f is the transducer resonance center frequency. Looking at Figure 1.16, it results evident that shorter pulses (i.e. higher frequency transducers) are desired to allow echoes from two targets in the axial direction be separated or correctly resolved. Once knowing the pulse duration, the axial resolution can be determined using the following expression: 𝑅 𝑎𝑥𝑖𝑎𝑙 = 𝑁 𝑐 𝜆 𝑙 2 (1.20) 26 where λl is the sound wavelength in the loading medium. Alternatively, the axial resolution can be defined as (Foster et al., 2002): 𝑅 𝑎𝑥𝑖𝑎𝑙 = 𝑐 𝑙 2𝑓𝐵𝑊 (1.21) where cl is the acoustic wave velocity in the loading medium, f is the transducer resonance center frequency, and BW is the -6 dB fractional bandwidth of the received echo signal. Looking at equations 1.19 thru 1.21, it becomes evident that the axial resolution can be improved by increasing the resonance center frequency and the bandwidth of an ultrasound transducer. Lateral resolution refers to the ability to distinguish two closely spaced reflectors that are positioned perpendicular to the axis of the ultrasound beam. Lateral resolution is most closely related to the transducer beam width (Zagzebski, 1996) as shown below in Figure 1.17. If the reflectors are separated more laterally than the beam width, then they are correctly resolved. Conversely, if the reflectors are closer than the beam width, they cannot be correctly resolved and their images merge together on the display. Figure 1.17. Lateral resolution refers to how closely positioned two reflectors can be perpendicular to the beam axis and be seen as separate objects on an ultrasound image. 27 Lateral resolution can be determined using the following expression (Cobbold, 2007): 𝑅 𝑙𝑎𝑡𝑒𝑟𝑎𝑙 = 𝑓 # 𝜆 𝑙 (1.22) where the f-number, f# = zf / D (zf : focal distance, D: transducer aperture) and λl is the sound wavelength in the loading medium. From these expressions, it is clear that the lateral resolution can be improved by increasing the resonant center frequency and/or the aperture size of a transducer. Figure 1.18 shows the sound field of a circular piston transducer. As it can be seen, the sound field is divided into two zones: the near field, or Fresnel zone, and the far field, or Fraunhofer zone. The near field is the region directly in front of the transducer where the sound pressure amplitude goes through a series of maxima and minima and ends at the last maximum: zo or transition point. The sound field remains well collimated in the near field, even narrowing down a bit for axial points approaching z0. The far field is the region beyond z0 where the sound pressure amplitude decreases approximately according to 1/z. Conversely, the sound field in the far field diverges with increasing distance from the transducer. Figure 1.18. Sound pressure field as a function of distance (z) for a circular piston transducer under continuous wave (CW) excitation. (Courtesy of Olympus, Inc.) 28 The length of the near field depends on the aperture of the transducer element (e.g. diameter, d) and on the ultrasound wavelength, λ. The length of the near field is given by: 𝑧 𝑜 = 𝑑 2 4𝜆 (1.23) The divergence angle (ϴ) of the far field also depends on the aperture of the transducer element and on the ultrasound wavelength: sin𝛳 = 1.2𝜆 𝑑 (1.24) As shown in Figure 1.13, single element transducers can be focused either by an acoustic lens attached to the planar piezoelectric element or by using an element that is curved. Focusing has the effect of narrowing the sound beam profile (i.e. better lateral resolution) and increasing the amplitude of echoes from reflectors over a limited axial range in comparison to an equivalent but unfocused transducer. The general principles of focusing are identical to those in optics (Shung, 2006). When choosing an acoustic lens, a convex lens is often preferred over a concave one in biomedical ultrasonic imaging because it conforms better to the shape of the body curvature. Additionally, the sound velocity in a convex lens is less than that in the medium into which the beam is emitted. The focal length (zf) of a lens is given by the following expression: 𝑧 𝑓 = 𝑅 𝑐 1− 𝑐 2 𝑐 1 (1.25) where Rc is the radius of curvature, c1 is the sound velocity in the lens, and c2 is the sound velocity in the medium. 29 As it can be inferred from the aforementioned concepts, in determining design parameters of ultrasound transducers such as the aperture size, focal depth, center frequency and others, a compromise has to be made between resolution requirements and the ability to obtain satisfactory sound beam penetration in order to image all the tissues of interest. At present, high-frequency single-element transducers are still widely used in research and clinical applications, such as monitoring tissue inflammation and responses to drug treatments (Chen et al., 2014), studying scattering from non-nucleated biological specimens (Falou et al., 2008), intravascular imaging (Yoon et al., 2015), B-scan ultrasound backscatter microscopes for dermatology imaging (Turnbull et al., 1995), and three-dimensional imaging of the anterior segment of the ocular globe (Silverman et al., 1995). 1.4 Ultrasound Arrays At present, most medical ultrasound imaging systems include array transducers rather than single-element transducers. Array transducers allow electronic steering of the ultrasound beam therefore circumventing the need to manually steer it in order to produce two-dimensional images as it is the case with single-element transducers. An array transducer consists of a group of closely spaced piezoelectric elements, each one with its own electrical connection to the ultrasound system. As a result, the elements can be excited individually or in groups to produce ultrasound beams; likewise, returning echo signals can also be detected by individual elements or by a group of elements to be afterwards amplified and combined into one signal. Figure 1.19 and Table III show and summarize the characteristics of the most common types of array transducers commercially available for clinical applications: (a) Linear sequential, 30 (b) curvilinear sequential, (c) linear phased, (d) 1.5D phased and (e) 2D phased arrays (Shung, 2006; Prince & Links, 2014). Figure 1.19. Array transducers commercially available for clinical applications: (a) Linear sequential, (b) curvilinear sequential, (c) linear phased, (d) 1.5D phased and (e) 2D phased arrays. (Prince & Links, 2014) 31 Table III. Main characteristics of array transducers. (Shung, 2006; Prince & Links, 2014) Array type Main characteristics (a) Linear sequential 1D - Usually 1 cm wide and 10 to 15 cm long - 128 to 256 elements, sometimes up to 512 elements - Typically 32 or more elements fired as a group in succession - Rectangular image field - Lateral scan only in the azimuthal plane - Elevation aperture is fixed - Focusing in the elevation plane is achieved with a lens - Azimuth aperture apodization to reduce grating lobes (b) Curvilinear sequential 1D - Similar to linear sequential but with a wider field of view - Pie-shaped image field (c) Linear phased 1D - Usually 1 cm wide and 1 to 3 cm long - Up to 128 elements - Elements fired separately - Electronic beam steering and dynamic focusing in the azimuth - Pie-shaped image field - Elevation aperture is fixed - Focusing in the elevation plane is achieved with a lens - Azimuth aperture apodization to reduce grating lobes (d) 1.5D phased - Additional elements in the elevation direction increase the number of electronic channels and array fabrication complexity - Number of elements in the elevation direction (5 to 7) is much less than the number in the azimuth direction (100 to 200) - Complete columns of elements fired separately - Electronic beam steering and dynamic focusing in the azimuth - Provides limited focusing capability in the elevation plane - Grating lobes in the elevation plane as a result of the small number of elements in the elevation direction - Azimuth aperture apodization to reduce grating lobes - Elevation aperture apodization and focusing are dynamically variable but symmetrical with respect to the center of the array (e) 2D phased - Additional elements in the elevation direction increase the number of electronic channels and array fabrication complexity - Number of elements in the elevation direction (40 to 60) is the same number in the azimuth direction (40 to 60) - Complete columns or rows of elements fired separately - Electronic beam steering and dynamic focusing in both azimuth and elevation directions - Azimuth aperture apodization to reduce grating lobes - Elevation aperture apodization and focusing are dynamically variable and can be non-symmetrical respect to the array center 32 Due to its versatility to electronically steer and focus the ultrasound beam, the phased array technology is currently the most useful. Referring to Figure 1.20, the time differences due to steering and focusing are expressed, respectively, by the first and the second terms on the right-hand side of the following equation (Shung, 2006): ∆𝑡 𝑛 = ∆𝑟 𝑛 𝑐 = 𝑥 𝑛 sin 𝜙 𝑥 𝑐 + 𝑥 𝑛 2 2𝑐𝑟 (1.26) where the difference in path length between the center element and element number, n, is defined as Δrn=r – rn, at a point P(r,ϕx). Figure 1.20. Coordinate system depicting the difference in path length between the center element of a linear phased array and the nth element. (Shung, 2006) On this basis, the pulse exciting the central element should be delayed by a time interval Δtn relative to the pulse that excites the element n if the pulses are required to arrive at point P at 33 the same time. In other words, the ultrasonic beam can be electronically focused and steered by properly delaying the signals going to the elements for transmission or arriving at the elements for reception, as shown in Figure 1.21. Figure 1.21. The echoes returned from focal point P can be made to arrive simultaneously by applying delays to the echoes detected at the elements of the phased array. (Courtesy of Analog Devices, Inc.) Phased arrays allow dynamic focusing in transmission and in reception. However, multiple transmissions of pulses are needed for dynamic focusing during transmission, slowing down the frame rate. Transmission dynamic focusing is usually done in discrete zones, whereas reception dynamic focusing can be done in many more zones or almost continuously, as shown in Figure 1.22 (Shung, 2006). Figure 1.22. Discrete zone dynamic focusing is used for transmission, whereas almost continuous dynamic focusing may be used for reception in phased arrays. (Shung, 2006) 34 After all data are acquired, a composite image is formed, taking only the data from the zones where the beam is focused. To maintain the beam width throughout the depth of view, modern scanners use dynamic aperture apodization (shading of the array elements). By using fewer elements during echo reception from shallow reflectors, the effective beam pattern is narrow; whereas the active aperture is expanded as echoes arrive from deeper structures, therefore keeping the effective beam narrow at all depths (e.g. D1, D2, and D3) and the lateral resolution nearly constant over the entire imaged region as shown in Figure 1.23 (Maslak, 1985; Shung, 2006). Figure 1.23. Dynamic aperture apodization (shading of the array elements) can be implemented to maintain the ultrasound beam width narrow and constant at all imaging depths. (Shung, 2006) Another form of apodization consists in amplitude apodization, which is implemented during pulse transmission for the reduction of side lobes. This is a process whereby the amplitude of the driving signal is not uniform over the whole surface of the piezoelectric element, but is made to get progressively weaker from the center of the element outward. An array may be apodized by varying the amplitude of the excitation pulse applied to various 35 elements when a sound beam is formed, with the innermost elements getting a greater excitation than the outer elements (Figure 1.24). Figure 1.24. The amplitude of the input voltage signal can be uniform (left) or varied (right) across a transducer array aperture. An amplitude apodized signal is preferred over its uniform counterpart to help reduce the appearance of spurious side lobes. Side lobes and grating lobes are two closely related phenomena resulting from sound energy that spreads out from the transducer at angles other than the primary beam path. Side lobes can be present in both single-element transducers and linear-array type transducers, but grating lobes are unique to linear-array type transducers. Figure 1.25 makes a clear distinction between side lobes and grating lobes. These unwanted sound energy beam paths can reflect off surfaces in the specimen under inspection and cause spurious indications on an image. The amplitude of grating lobes is affected by the pitch size, number of elements, frequency, and bandwidth. Figure 1.26 compares two beam profiles produced by different transducers with approximately the same aperture size. The beam on the left has an excellent shape with all of its energy concentrated in the main lobe, while the beam on the right presents two spurious lobes at approximately ±30 off the main lobe. 36 Figure 1.25. Distinction between grating and side lobes. Grating lobes are indeed high side lobes that occur at certain angles due to constructive interference. In this case grating lobes are produced at ±40 by a linear array with an undesirable pitch. (Shung, 2006) Figure 1.26. Comparison of beam profiles without (left) and with (right) spurious lobes. The beam at left is generated by six elements at 0.4 mm pitch and the beam at right by three elements at 1.0 mm pitch. (Courtesy of Olympus, Inc.) 37 Figure 1.27 further clarifies why artifacts are produced with grating lobes. Considering the beam profile of Figure 1.26 with grating lobes at ±30 off the main beam, and supposing the transducer is steered ±40 while imaging a wire point target, upon image reconstruction the user will see the image of the wire point target produced by the main lobe echo at 0 and two other spurious wire point targets at ±30 produced by echoes from the grating lobes. This becomes of crucial importance in situations such as tumor imaging. Because of grating lobes, a tumor could be seen larger or affecting more areas than what it really is, thus resulting in incorrect diagnosis. Figure 1.27. Two false point targets are produced when steering a linear phased array transducer with spurious lobes at ±30 off the main lobe. The third picture on the right shows an example with a phantom composed of multiple wire point targets. (Lockwood et al., 1996) As it can be deducted from the following equation that relates the angle (ϕg) at which grating lobes occur to the wavelength (λ) and the pitch (g), where n is an integer = ±1, ±2… 𝜙 𝑔 = sin −1 ( 𝑛𝜆 𝑔 ) (1.27) for grating lobes to occur at angles greater than ±90 off the main lobe (grating lobe suppression) the pitch must be smaller than half wavelength (fully-sampled array). As pitch is reduced, the grating lobes will move further away from the main lobe. Other ways to suppress grating lobes 38 include sub-dicing of elements, or randomizing the spacing between elements (Lockwood & Foster, 1995). There are a few simple design rules for linear sequential and linear phased arrays, all of which are summarized in Figure 1.28 and Table IV. Figure 1.28. Design parameters of linear sequential and linear phased arrays. Table IV. Design rules of linear sequential and linear phased arrays. (Shung, 2006) Array type Pitch g Element width b Element thickness L Ratio of element width to element thickness b/L Height h Crosstalk Linear sequential 0.75λl - 2λl > 0.5λl ≈ 0.5λp < 0.6 ≥ 10L < -30 dB Linear phased ≤ 0.5λl ≤ 0.5λl ≈ 0.5λp < 0.6 ≥ 10L < -35 dB λ l is the ultrasound wavelength in the loading medium; λ p is the ultrasound wavelength in the piezoelectric material. As previously stated in section 1.3 Ultrasound Transducers, spatial resolution is a measure of how closely positioned two reflectors can be and still be distinguished from each other on an ultrasound image. Lateral resolution indicates how close the reflectors can be along a 39 line perpendicular to the ultrasound beam and remain distinguishable; depending on the beam width in the imaging plane. Axial resolution indicates how close the reflectors can be along the ultrasound beam axis and still be resolved; depending on the ultrasound pulse duration. A third measure of spatial resolution, the elevational resolution, also known as the slice thickness (Figure 1.29), is equally important. Slice thickness is measured in a direction perpendicular to the image plane and determines the thickness of the section of tissue that contributes to echoes visualized on the ultrasound image. Slice thickness depends on the beam size, analogous to lateral resolution; however, it is the size of the beam perpendicular to the image plane, rather than in the image plane as in lateral resolution, that establishes slice thickness. Figure 1.29. An ultrasound beam has three measurements of spatial resolution: Axial, lateral, and slice thickness resolution. (Penninck & d’Anjou, 2015) For most linear sequential, curvilinear sequential, and linear phased arrays the slice thickness and the lateral resolution are very different. Previously discussed electronic focusing 40 only reduces the in-plane width of the beam; therefore improving the lateral resolution, but it has no effect on slice thickness. Beam width in the slice thickness direction is determined by a fixed focal length lens attached to the entire array (Figure 1.30). The slice thickness usually is approximately the size of the scanhead up close to the array, narrows down at the lens focal distance, and then broadens out at points beyond the lens focal distance (Figure 1.29). At present, the slice thickness is considered the worst measure of spatial resolution for linear sequential, curvilinear sequential, and linear phased array transducers (Zagzebski et al., 1982). Figure 1.30. Detailed construction isometric view of a linear array transducer where the plastic housing has been partially cut out to reveal the internal components. For the correct diagnosis of a medical condition, it is of crucial importance for the clinician to bear in mind that because of slice thickness, ultrasound images are not depicting an anatomical plane in the body, but a volume of tissue (Figure 1.31). The volume of tissue is thick up close to the transducer scanhead, narrower at the focal depth of the elevational focus, and thicker again at distances beyond the slice thickness focus. 41 Figure 1.31. Schematic depiction of the volume of tissue imaged during a B-mode scan done with a phased array. (Zagzebski, 1996) As it can be inferred from the aforementioned concepts, the design of ultrasound arrays represents a more challenging task than that corresponding to the design of single-element ultrasound transducers. Apart from the main parameters such as aperture size, focal depth, center frequency, lateral and axial resolutions; ultrasound arrays require additionally determining the number of elements, their pitch and geometry, if they are fired together as a group or separately, the maximum angle to steer the beam in order to avoid grating lobes (phased arrays), whether implementing dynamic aperture or amplitude apodization, or a combination of both; if a lens is needed for fixed focusing in the elevation plane (1D arrays) or if elevation focusing must be done dynamically (1.5D and 2D arrays), among several other parameters requiring a compromise to be made in order to obtain satisfactory spatial resolution and sound beam penetration to image all the tissues of interest. At present, ultrasound array transducers are widely used in diverse clinical applications: In gastrointestinal imaging, one application employs ultrasound arrays during the initial evaluation of patients with abdominal pain for the early detection of bowel disease (Chaubal et al., 2006); in cardiovascular imaging, one of the latest reported applications employs ultrasound arrays as screening tools for the prediction of cardiovascular implantable electronic device lead 42 fibrosis (Yakish et al., 2015); in the constantly growing field of obstetrics and gynecology (OB/GYN), the recently launched ACUSON S3000 TM features a variety of ultrasound array probes that enable the acquisition of excellent resolution images at depth even on third trimester patients (Siemens Healthcare, 2015); in radiology, ultrasound arrays are being used for quantitative shear wave elastography, a method of obtaining quantitative tissue elasticity data during breast ultrasound examinations (Evans et al., 2010); and in ophthalmic imaging, an ultrasound array has been used to cover the whole eye within a single image, creating multiple focal zones with dynamic aperture and with the capability of implementing Doppler or color flow mapping (Kim et al., 2007). 1.5 Objectives of Research The goal of this research is to provide array transducer solutions for: 1. A miniaturized robotic surgery probe for real-time fusion of ultrasound and endomicroscopy during Transanal Endoscopic Microsurgery (TEM). 2. A forward-looking (FL) catheter probe for intravascular ultrasound (IVUS) imaging and intraluminal vessel navigation applications. Chapter 2 discusses the array transducer for the miniaturized robotic surgery probe. The discussion starts with an overview of colorectal cancer, TEM, and explains the clinical need for the inclusion of the array transducer to the probe. Afterwards, the discussion centers on the array transducer design following the design rules previously presented in section 1.4 Ultrasound Arrays, and the final construction design is described using computer-aided design (CAD) software. The discussion continues with simulation modeling using a finite-element analysis 43 (FEA) software package to move on to the fabrication procedure and finish with the results obtained from performance characterization and imaging evaluation testing. Chapter 3 discusses the array transducer for the FL-IVUS catheter probe. The discussion starts with an overview of peripheral arterial disease (PAD), angioplasty procedures, and explains the clinical need for an array transducer with FL capability. Similarly to the structure employed in Chapter 2, the discussion afterwards continues with the array transducer design, FEA simulation modeling, fabrication procedure, and performance characterization and imaging evaluation testing. Chapter 4 summarizes the procedures and results of the two array transducer probes presented in this dissertation. 44 CHAPTER 2 MINIATURE LINEAR PHASED ARRAY FOR COLORECTAL ROBOTIC SURGERY This chapter describes the design, fabrication, and testing of a miniature side-looking (SL) linear phased array probe for Endorectal Ultrasound (ERUS) imaging during colorectal robotic surgery procedures. The clinical problem and current standard of care are presented first in order to arrive to the clinical need this probe is intended to satisfy. The array features 64 piezoelectric elements separated by non-conductive epoxy kerfs, yielding a total active aperture of 3.2 mm in the azimuth direction and 1.8 mm in the elevation direction, with an elevation natural focal depth of 8.1 mm. The array includes non-conductive epoxy backing and two front matching layers. A custom flexible circuit connects the array piezoelectric elements to a bundle of 64 individual micro-coaxial cables (48 AWG) enclosed within a 1.5-m long 10F catheter and terminated with a Verasonics Backshell Connector. 2.1 Clinical Problem: Colorectal Cancer (CRC) Colorectal cancer (CRC) is the development of cancer in the colon or rectum, resulting from the abnormal growth of cells that may invade or spread (metastasis) to other parts of the body (Figure 2.1). The signs and symptoms depend on the location of the tumor in the bowel, and whether it has metastasized. These include blood in the stool, a change in bowel movements, decrease in stool caliber, loss of appetite and weight, nausea, vomiting, and feeling tired continuously (Yamada et al., 2011). 45 Figure 2.1. Precancerous polyps (adenomas) grow within the lining of the colon and rectum. Over time, a few of these polyps can become cancerous (adenocarcinomas) and may penetrate the colon and rectum walls, spreading to lymph nodes along with other organs. (Courtesy of Blausen Medical, Inc.) CRC is the third most diagnosed cancer in men, next to prostate and lung cancer. In women it is the third most diagnosed cancer, next to breast and lung cancer. In the United States, according to the American Cancer Society (ACS) and the Centers for Disease Control and Prevention (CDC) there will be 134,784 new cases of CRC in 2015, being the second leading cause of cancer related deaths of men and women together (Figure 2.2). Figure 2.2. Cancer statistics for men and women in the United States. Colorectal cancer is the third most diagnosed cancer in both men and women, and the second leading cause of cancer related deaths of men and women together. (ACS and CDC, 2015) 46 2.2 Current Treatment: Transanal Endoscopic Microsurgery (TEM) Transanal Endoscopic Microsurgery (TEM) is a minimally invasive procedure for the resection of colorectal lesions that employs a natural orifice approach (via the anus) rather than the more traditional abdominal and open approaches (De Graaf et al., 2002). TEM is utilized for the excision of rectal tumors, early stage rectal cancers, and benign rectal polyps. Through this procedure, patients benefit from the following: No abdominal incision, reduction of bleeding, less risk of infection, reduced post-surgery bowel obstructions, and shorter hospital stay with faster recovery (Demartines et al., 2001). Among the diverse pieces of equipment employed during a TEM procedure, colorectal surgeons mainly rely on specialized rectoscopes and endoscopic instruments in order to improve visualization and dexterity in the rectum, allowing for full-thickness local resection beyond the reach of traditional surgical instruments (Figure 2.3). Figure 2.3. Resection of colorectal tumor via TEM. (Courtesy of European School of Oncology) 47 Currently, TEM procedures rely on preoperative endoscopic biopsies for tissue characterization and Magnetic Resonance Imaging (MRI) or Endorectal Ultrasound (ERUS) for staging the malignancies (Zacharakis et al., 2007). CRC staging is done according to the TNM Staging System (Figure 2.4) developed and maintained by the Union for International Cancer Control (UICC) and the American Joint Committee on Cancer (AJCC). Overall, T describes the size of the original tumor and whether it has invaded nearby tissue, N describes nearby lymph nodes being affected, and M describes distant metastasis. Figure 2.4. The T stages of CRC. (Courtesy of Cancer Research United Kingdom) Endoscopic biopsies are performed with Colonoscopy, which employs a flexible fiber optic endoscope for evaluation of the entire colonic mucosa with therapeutic capability of resecting detected malignancies (Figure 2.5). Colonoscopy removes all detected polyps, regardless of histology type (adenomas or adenocarcinomas). However, the polyps vary in size and those under 5 mm are not detected endoscopically; furthermore, the CRC diagnosis can be confirmed only after the biopsy has been examined (Stevenson et al., 1992). 48 Figure 2.5. Colonoscopy is an endoscopic technique used to screen individuals for CRC. During a Colonoscopy procedure, a flexible fiber optic endoscope is introduced through the anus till the start of the colon. (Courtesy of Medical Billing and Coding Online) The standard phased array MRI produces good quality images with good contrast resolution and a relatively large field of view, so it is the modality of choice for the preoperative staging of a rectal tumor. MRI is reliable for tumor assessment and its locoregional extension, for identifying recurrence, and for planning radiation therapy. The disadvantage of MRI is its incapability for evaluating nodal metastases. Additionally, MRI may not be the examination of choice for every patient. Patients with contraindications to MRI (e.g. implantable pacemakers) or unable to tolerate MRI (e.g. due to claustrophobia) would preferably undergo preoperative imaging with Computed Tomography (CT), with the concomitant exposure to ionizing radiation. Furthermore, motion related imaging artifacts could severely affect the diagnostic quality of both MRI and CT for patients who are unable to hold their breath for longer than 20 seconds (Ignatov et al., 2014). 49 Endorectal Ultrasound (ERUS) is currently the most widely used and effective diagnostic modality in the assessment of CRC overall (Figure 2.6). Its accuracy in numerous trials ranges from 80 to 95% for T-staging and 70 to 75% for N-staging, levels that are slightly higher than the respective 75 to 85% and 60 to 70% observed for MRI (Edelman & Weiser, 2008). ERUS does not expose the patient and clinical staff to ionizing radiation, and the patient is not kept immobilized (Pratt et al., 2012). Additionally, ERUS does not require the patient to be enclosed inside a chamber and there are practically no contraindications for its use (e.g. patients with implantable pacemakers can undergo ERUS). Figure 2.6. Endorectal Ultrasound (ERUS) also known as Transrectal Ultrasound (TRUS). An ultrasound probe is inserted into the rectum for examination of the walls and tissue lying underneath. (Courtesy of WebMD, LLC) In view of the diverse imaging modalities currently available, there has been considerable debate about which one represents the best option for the examination of CRC offering both detection and characterization. However, what results more critical is the high recurrence rate, 50 approximately 37.5%, due to incomplete excisions derived from the lack of intraoperative image guidance with real-time histological data during TEM procedures (McCloud et al., 2006). 2.3 Clinical Need To summarize the clinical need, colorectal surgeons need a means that allows them to execute TEM procedures with both intraoperative image guidance and real-time histological data to assess the surgical margins intraoperatively and adjust the procedure accordingly. To fulfill this need there have been several attempts combining multiple imaging modalities in order to provide a more detailed assessment of the tissue. One example integrates US, Optical Coherence Tomography (OCT), and photoacoustic tomography into a single instrument, providing information on a similar plane at different depths and resolutions (Yang et al., 2011). Another example being developed by the Hamlyn Centre for Robotic Surgery at Imperial College London and our group, the Resource Center for Medical Ultrasonic Transducer Technology at the University of Southern California, combines Endomicroscopy and ERUS mounted on a novel robotic instrument to provide diagnostically complementary information for the intraoperative assessment of CRC surgical margins during TEM procedures (Figure 2.7). Endomicroscopy provides cellular level information at a superficial level whereas ERUS provides depth resolved information at a macroscopic level. The ability to combine these modalities can assist the assessment of surgical margins through the use of Endomicroscopy, whereas the overall three-dimensional structure of the cancerous tissue can be inferred from the ERUS images, along with the thickness of the rectum, assisting in preventing perforation of the 51 lumen (Dwyer et al., 2015). Figure 2.7. Robotic surgery probe for real-time intraoperative fusion of ERUS and Endomicroscopy during TEM procedures. (Courtesy of the Hamlyn Centre for Robotic Surgery, Imperial College London) The following sections of the current chapter present the miniature linear phased array developed by our group for the robotic surgery probe being developed by the Hamlyn Centre for Robotic Surgery. 2.4 Array Transducer Design The array design incorporates the required features as informed by the Hamlyn Centre for Robotic Surgery: Array type, aperture size, frequency, orientation, and catheter packaging size. A side-looking (SL), linear phased array is needed in order to image the colon and rectum walls, as well as the tissue lying underneath. In practice, the colorectal surgeon would insert the array transducer catheter into the robotic surgery probe until the array aperture lies right under the machined window of the frame for the US array (Figure 2.7). Once in place, the array is capable of producing real-time images from inside the colon and the rectum that can also be 52 reconstructed to render a three-dimensional structure from which cancerous tissue can be inferred, along with the thickness of the rectum, assisting in preventing perforation of the lumen. The SL-ERUS linear phased array catheter was modeled via the computer-aided design (CAD) software SolidWorks (Dassault Systèmes SolidWorks Corp., Waltham, MA). As shown in Figure 2.8, the array acoustic stack is enclosed within a 10F catheter having an internal diameter of 3 mm together with a flexible circuit that clamps the 2-2 composite on its proximal and distal ends to provide ground connection. The design of the flexible circuit allows connecting and fitting a bundle of individual micro-coaxial cables (48 AWG) inside of the catheter to excite each one of the piezoelectric elements. Figure 2.8. CAD model of SL-ERUS linear phased array enclosed in protective 10F catheter. The array must have significantly high center frequency and bandwidth in order to achieve the required resolution and provide enough depth resolved information. Currently, clinical ERUS images are usually obtained using an ultrasound frequency of 5 to 15 MHz, depending on which part of the rectum is being examined. Higher frequencies provide better 53 resolution of the sphincter muscles and the rectal wall layers, whereas pararectal tissue and lymph nodes are more accurately assessed using lower frequencies (Starck et al., 2003). 2.4.1 Array Acoustic Stack The array center frequency was chosen to be 15 MHz to achieve axial and lateral resolutions necessary to reliably visualize the full thickness of the rectal walls and assist in the overall three-dimensional identification of the cancerous tissue as well as to prevent perforation of the lumen. The final array geometry and materials were determined considering the concepts and design rules presented in sections 1.3 Ultrasound Transducers and 1.4 Ultrasound Arrays, as well as simulation results and fabrication process limitations. Lead magnesium niobate-lead titanate (PMN-PT) (CTS Corporation, Bolingbrook, IL) single crystal was chosen as the piezoelectric material due to its high dielectric constant and low dielectric loss, which make it ideal for high-sensitivity transducers with small-aperture size (Chen et al., 2013). The array 2-2 composite consists of sixty-four 37- m-wide piezoelectric elements separated by 13- m-wide kerfs filled with non-conductive epoxy (Epo-Tek 301) (Epoxy Technology, Inc., Billerica, MA) to yield an array with a 50- m-wide pitch (0.5 water pitch for the suppression of undesirable grating lobes during beam steering), a total active azimuth aperture of 3.2 mm, and a total active elevation aperture of 1.8 mm; features that represent an important and considerable progress in the area of miniature SL catheter array transducers where most current designs employ two times larger pitches (~100 m) with concomitant larger element-kerf sizes and larger imaging apertures (Stephens et al., 2008). The array acoustic stack includes a backing layer of non- conductive epoxy (Epo-Tek 301) mixed with 43.2 wt% liquid plasticizer (LP-3) (Structure Probe, Inc., West Chester, PA) and two front matching layers: 1. Conductive silver-loaded epoxy 54 consisting of 2,3- m silver powder (Aldrich Chemicals Co., St. Louis, MO) with mixed epoxy (Insulcast 501 + Insulcure 9) (ITW Polymers Coatings North America, Montgomeryville, PA) and 2. Vapor deposited polymer (Parylene C) (Specialty Coating Systems, Inc., Indianapolis, IN). A rendering of the final array design, its design parameters, and material properties are shown next in Figure 2.9, Table V, and Table VI, respectively. Figure 2.9. CAD exploded view of array acoustic stack including backing, 2-2 array composite, 1 st matching layer (silver epoxy), and 2 nd matching layer (vapor deposited polymer) as well as zoomed view of separate layers and the final assembled array unit. The flexible circuit is included to show its positioning between the 2-2 array composite and the backing. Table V. SL-ERUS linear phased array design parameters. Design center frequency 15 MHz Number of elements 64 Composite configuration 2-2, linear Pitch (0.5 water) 50 m Element width 37 m Kerf width 13 m Elevation aperture 1.8 mm Azimuth aperture 3.2 mm Elevation focal distance (natural) 8.1 mm Piezoelectric material PMN-PT 53 m thick Kerf filler material Non-conductive epoxy (Epo-Tek 301) 53 m thick 55 Backing material Non-conductive epoxy (Epo-Tek 301) with 43.2 wt% liquid plasticizer (LP-3) 2 mm thick 1 st Matching layer material Conductive epoxy (2,3- m silver epoxy) 45 m thick 2 nd Matching layer material Vapor deposited polymer (Parylene C) 30 m thick Table VI. Material properties of the SL-ERUS linear phased array acoustic stack. Material PMN-30%PT single crystal 1 Epo-Tek 301 2 Epo-Tek 301 + LP-3 (43.2 wt%) 3 2,3- m silver epoxy 4 Parylene C 5 Density, ρ (kg/m 3 ) 7,800 1,150 1,230 3,860 1,100 Acoustic longitudinal velocity, c (m/sec) 4,600 2,650 2,450 1,900 2,350 Acoustic impedance, Z (MRayl) 35.88 3.05 3.01 7.33 2.59 Piezoelectric strain constant, d33 (C/N) 1,500 x 10 -12 - - - - Free dielectric constant, K к ~5,000 - - - - Clamped dielectric constant, K ε 800 - - - - Dielectric loss tangent, tanδe 0.005 - - - - Electromechanical coupling coefficient, kt 0.58 - - - - 1. CTS Corporation, Bolingbrook, IL 2. Epoxy Technology, Inc., Billerica, MA 3,4. Resource Center for Medical Ultrasonic Transducer Technology at the University of Southern California 5. Specialty Coating Systems, Inc., Indianapolis, IN 56 2.4.2 Electrical Interconnect – Flexible Printed Circuit One critical challenge of fabricating miniaturized high-frequency catheter arrays is creating an electrical interconnect solution. For the array under discussion, the element width and pitch, together with the number of elements and the available space inside of a 10F catheter (3mm internal diameter), required designing a high-density flexible circuit that pushes the limits of current technology for flexible printed circuitry production in terms of electrical trace width and pitch. Figure 2.10 shows the CAD schematics (AutoCAD, Autodesk, Inc., San Rafael, CA) of the flexible circuit design that allows connecting and fitting a bundle of 64 individual 48 AWG micro-coaxial cables inside of the 10F catheter to excite each one of the 64 piezoelectric elements. The flexible circuit features sixty-four 30- m-wide signal traces with a 50- m-wide pitch on the 2-2 array composite area, divided into two groups of 32 signal traces each. Each group further divides into three branches of 10 (center tab) or 11 (side tabs) signal traces, each of which will then be connected to an individual 48 AWG micro-coaxial cable. Connection pads are 100 m wide and their pitch is 180 μm. Figure 2.10. CAD schematics of flexible printed circuit design. 57 2.4.3 Electrical Interconnect – Cabling The micro-coaxial bundle features 64 individual 48 AWG micro-coaxial cables (Hitachi Metals, Ltd., Tokyo, Japan), whose properties are shown below in Figure 2.11. 3. PFA = Perfluoroalkoxy Figure 2.11. High-capacitance 48 AWG micro-coaxial cable. (Courtesy of Hitachi Metals, Ltd.) In order to determine the optimal micro-coaxial cable length to be used, scattering parameters (S-Parameters) were evaluated to achieve the best performance in terms of reflection and transmission. S-Parameters are employed to represent high-frequency networks in accordance to direct measurements of reflected and transmitted waves. They provide a complete description of the network regarding how the incident waves are related to the reflected and transmitted waves (Figure 2.12). S21 S11 S12 S22 2 Port Network + - V1 V1+ V1- + - V2 V2+ V2- Figure 2.12. 2-Port Network representation of a micro-coaxial cable for estimating its scattering parameters: Reflection and Transmission Coefficients. 58 S-Parameters are defined as (Pozar, 2012): 2 1 22 12 21 11 2 1 V V S S S S V V (2.1) 1 1 11 V V S where 0 2 V It defines the reflection at port 1 due to impedance mismatch. 1 2 21 V V S where 0 2 V It defines the gain (loss) of system from port 1 to port 2. 2 1 12 V V S where 0 1 V It defines the gain (loss) of system from port 2 to port 1. 2 2 22 V V S where 0 1 V It defines the reflection at port 2 due to impedance mismatch. The S-Parameters of the 48 AWG micro-coaxial cable were measured at three different lengths (1.5, 2, and 2.4 m) with an Agilent E5072A Network Analyzer (Agilent Technologies, Santa Clara, CA). Due to reciprocity of the system, S11 = S22 and S21 = S12. Figures 2.13 and 2.14 show the effect of micro-coaxial cable length on the S-Parameters. Figure 2.13. Reflection coefficient of the 48 AWG micro-coaxial cable at 1.5, 2.0, and 2.4 m. 59 Figure 2.13 shows that at 15 MHz the reflection coefficient of the 48 AWG micro-coaxial cable increases with increasing cable length. It depicts that 1.5 m cable length generates less reflection at the input and output; therefore having less oscillatory effect on the transmitted and received signal. Figure 2.14. Transmission coefficient of the 48 AWG micro-coaxial cable at 1.5, 2.0, and 2.4 m. Figure 2.14 shows that at 15 MHz the transmission coefficient of the 48 AWG micro- coaxial cable decreases with increasing cable length. In other words, the loss of transmitted signal increases with increasing cable length. Therefore 1.5 m cable length delivers higher power to the drive the array transducer. In conclusion, both S11 and S21 indicate that 1.5 m is the most efficient option of the three cable lengths tested and exerts the lowest attenuation on the signal. Therefore, we chose 1.5 m length for each individual 48 AWG micro-coaxial cable. 60 2.4.4 Electrical Interconnect – Verasonics Backshell Connector Kit The array is to be driven by a Verasonics Vantage 128 System (Verasonics, Inc., Kirkland, WA). In order to take advantage of the system capabilities and to avoid any possible incompatibility drawback, a Backshell Connector Kit (Verasonics, Inc., Kirkland, WA) was procured to connect the array to the system. The connector kit (Figure 2.15) includes: 4 micro- coax termination boards (plus 2 spares) of which only 2 are to be used by the array being presented; backshell left and right halves, 1 ZIF (zero-insertion force) connector PCB (printed circuit board), 1 Cannon DL-260 connector plug and locking handle, 2 tuning PCBs, and assembly documentation. Overall, to assemble the kit, the micro-coaxial cables must first be soldered to 2 termination boards, which handle 32 signal channels each. Additionally, should tuning components be necessary, these also have to be soldered to the tuning boards. The termination boards are then plugged into the tuning boards, and these, into the ZIF connector PCB. The resulting subassembly is then placed and fitted inside one backshell half and finally the other backshell half is secured in place with the included screws. A more detailed description of the Backshell Connector Kit assembly for the array under discussion is included ahead in section 2.6 Array Transducer Fabrication. 61 Figure 2.15. Verasonics Backshell Connector Kit. (Courtesy of Verasonics, Inc.) 2.4.5 Verasonics Imaging System The Verasonics Vantage 128 System, shown in Figure 2.16, counts with 128 transmit (Tx) and 128 receive (Rx) channels that are independently controlled, of which only 64 Tx / 64 Rx channels will be used by the array under discussion. On Tx mode, the system is to be operated at a programmable center frequency of 15 MHz with pulse amplitudes of 100 Vpp or lower. Even though the array has been designed with a 0.5λwater pitch, the system also counts with per-channel Tx apodization using pulse-width modulation to assist in the suppression of undesirable grating lobes during beam steering. On Rx mode, the system counts with real-time access to RF data for each individual channel should it be necessary (e.g. verify that an array element is indeed receiving returning echoes and generating corresponding electrical signals) and a programmable anti-aliasing filter. The technical specifications of the Verasonics Vantage 128 System are summarized below in Table VII. 62 Figure 2.16. Verasonics Vantage 128 System. Table VII. Verasonics Vantage 128 System technical specifications. Transmit (Tx) All transmit and receive channels are independently controlled Standard frequency range: 0.5 MHz to 20 MHz o (2 – 42 MHz with High Frequency configuration) o (100 KHz – 1700 KHz with Low Frequency configuration) Time delay resolution: 4.0 nsec. Programmable pulser amplitude, 3 to 190V pp Tri-state drive: + high voltage, – high voltage and ground Per-channel, programmable center frequency, pulse width (pulse duty cycle), pulse length, polarity and delay Per-channel transmit apodization using pulse width modulation Max burst length (without Extended Transmit option) of 20 cycles at transmit center frequency Power limit, single channel: up to 100 Watts peak, 8 Watts average (into 50 Ohms) Receive (Rx) Real-time access to RF data for each channel Standard frequency range: 0.5 – 27 MHz o (1 – 50 MHz with High Frequency configuration) o (100 KHz – 1700 KHz with Low Frequency configuration) Programmable anti-aliasing filter cutoff: 10, 15, 20 and 30 MHz, and in addition, 35 and 50 MHz with High Frequency configuration 14 bit A/D converters with programmable sample rate up to 62.5 MHz Two independent, user-programmable, symmetrical RF data digital (23 tap and 41 tap) 63 2.5 Array Transducer Simulation Modeling The final design of the SL-ERUS linear phased array was defined via the finite-element analysis (FEA) software PZFlex (Weidlinger Associates, Inc., Cupertino, CA), which is time- domain based and especially developed for modeling piezoelectric composite arrays for ultrasound imaging. FEA modeling is capable of providing an accurate prediction of the performance of a high-frequency array, therefore reducing the number of time-consuming prototype fabrication runs (Chen et al., 2014). The array performance was evaluated and optimized via simulations that included: 1. Pulse-echo response and electrical impedance, 2. Electrical and acoustical crosstalk, and 3. Pressure field and axial/lateral beam profiles. 2.5.1 Pulse-Echo Response and Electrical Impedance In order to perform the pulse-echo response and electrical impedance simulations, three files were created and executed: 1. Input (.flxinp) file for pulse transmission, 2. Input (.flxinp) file for echo reception, and 3. Review (.revinp) file for displaying results. 2.5.1a Input File for Pulse Transmission The input file for pulse transmission started with the specification of the array parameters listed in Table V, with all dimensions specified in meters. Additionally, in this section the following parameters were declared: Minimum sound wave velocity (m/sec) of the model with its loading medium, design center frequency (Hz), approximate element size and number of elements per wavelength at the highest frequency of interest (usually two times the design center frequency), thickness of the loading medium (m), and enough simulation time (sec) for two-way travel of pressure wave towards / back from the focal point. 64 The following step consisted in setting up the grid coordinates of key points of the array model. Even though the present discussion deals with 2D modeling, it is imperative to mention that PZFlex has the capability of performing 3D modeling of any kind of ultrasound transducers; naturally, at the expense of more complicated geometry construction and longer simulation times. Figure 2.17 shows the 2D model of the array with its respective x-y coordinates (symmetry boundary condition applied). 65 Figure 2.17. 2D model of the SL-ERUS linear phased array with its x-y coordinates (symmetry boundary condition applied). 66 The next step consisted in setting up the i-j indices to indicate node locations in the finite element 2D model. On this basis, the geometry was completely defined by associating the specified i-j indices with their respective x-y coordinates. Afterwards, specific materials were assigned to their respective regions in the model. These materials were read from a material (.matr) file, which must be included in the same folder as the input file under discussion. After assigning the materials, the model graphics were generated and reviewed. The following step consisted in applying boundary conditions to the four sides of the model. In this case, since the whole transducer is surrounded by water, three sides of the model (left, top, and right) were specified as absorbing (absr) boundaries. The bottom side of the model used symmetry (symm) as boundary condition to reduce geometry construction and simulation times. After specifying the boundary conditions, the driving function exciting the desired piezoelectric element(s) was defined. For the present array, a single-cycle sinusoid (pulsed wave excitation) with an amplitude of 100 Vpp at the design center frequency was chosen. Afterwards, it was necessary to define the electrical field within which the piezoelectric elements are located. For this purpose, an electrical window was defined to encompass all the 2-2 composite (from i5 to i6, and from j1 to j66. Looking at Figure 2.17, this happens from x5 to x6, and from y1 to y66). The type of transmitting electrode for the desired transmitting element was defined as a 50 Ω series resistor, and was located at the bottom of the piezoelectric element (i.e. at the interface between the dielectric polyimide and the piezoelectric material). Likewise, the common ground electrode was defined and located at the top of the piezoelectric material (i.e. at the interface between the piezoelectric material and the first matching layer). 67 The following step consisted in what could be considered the cornerstone of the input file for pulse transmission: Defining the Kirchoff extrapolation field and echo calculation. Kirchoff extrapolation is particularly useful with ultrasound transducer models having their focus at relatively large distances away from the transducer front. Such huge models would take a great amount of memory to run, especially considering discretization of a large loading medium. Instead, Kirchoff extrapolation creates a much smaller model extending only about five wavelengths past the pressure wave source (i.e. the piezoelectric element). It is of utmost importance to mention that Kirchoff extrapolation requires a homogeneous fluid loading medium (e.g. pure water) in order to be applied. The extrapolation is by no means valid when there is more than one kind of material or even when the same material is present in different physical states. For implementation of Kirchoff extrapolation, an imaginary box was defined around as much of the transducer as possible (see dotted line around transducer in Figure 2.17 – left, top, and right sides). It is important to mention that the sides of the box must not coincide with the solid edges of the model since these are used to define the boundary conditions. Next, a reference point from which to calculate the pressure gradient for extrapolation was defined. Such point must be located inside the transducer and centered as much as possible (i.e. x6, y1). Afterwards, a Kirchoff data plane was defined spanning the entire model front length, inside the fluid region, and within a couple of elements from the solid/fluid (i.e. 2 nd matching layer/water) interface. Next, for echo calculation, the coordinates corresponding to the position of the reflector (e.g. quartz crystal) were defined along with pulse duration and declaration that only half of the transducer was being simulated due to the symmetry boundary condition previously defined. The next step consisted in specifying which time histories to store in a history (.flxhst) file. For the array under discussion, echo time history together with electrode voltage and charge 68 histories (for the posterior calculation of electrical impedance) were stored. The last two steps of the input file for pulse transmission consisted in processing the model for execution and saving all the model variables into a symbol (.symb) file for their posterior use with both the input file for echo reception and the review file for displaying results. 2.5.1b Input File for Echo Reception The input file for echo reception was essentially the same as the input file for pulse transmission, just with the following modifications: In the boundary conditions section, the one corresponding to the right side of the model was changed from absorbing (absr) to free-field (ffld) boundary condition, so that the echo produced by the pressure wave hitting the reflector became the right-side boundary condition and therefore be perceived by the same desired piezoelectric element of the array transducer. The driving function exciting the desired piezoelectric element was removed. All other defined electrical parameters were left exactly the same with the only difference that the bottom electrode was defined as a 50 Ω shunt resistor (not series anymore). All Kirchoff extrapolation and echo command groups were removed. 2.5.1c Review File for Displaying Results The review file for displaying results started by making sure that the required input files (both pulse transmission and echo reception) had been previously run. For such purpose, a routine was programmed to inform whether the previous two input files needed to be run first or not. The next section consisted in reading all the data (variables and time histories) generated 69 from running the two input files. Once all the data had been read in, the electrical impedance of the desired piezoelectric element could be calculated using its respective voltage and charge time histories. Likewise, fast Fourier transform (FFT) could be computed on the time history of the voltage generated by the received echo on the receiving piezoelectric element, in order to produce the frequency spectrum of the echo. Finally, the last step consisted in plotting the obtained results, which could be done either within the PZFlex environment or externally via another software package. In our case, the results were saved and exported as .dat files for plotting them with MATLAB (The MathWorks, Inc., Natick, MA). Figure 2.18 shows the results obtained for one of the 64 piezoelectric elements of the SL-ERUS linear phased array (element 33). The left subfigure shows the voltage history generated by the received echo (μs vs. mV) and its frequency spectrum (MHz vs. dB), rendering a peak-to-peak sensitivity of 238.9 mV, an axial resolution of 80.25 μm (equation 1.20), a -6 dB center frequency of 14.1 MHz, and a -6 dB fractional bandwidth of 56.7%. The right subfigure shows the piezoelectric element electrical impedance magnitude (MHz vs. Ω) and phase angle (MHz vs. deg), rendering an electrical impedance magnitude of 123.8 Ω with a phase angle of -42.94° at 15 MHz and an electromechanical coupling coefficient (equation 1.13) of 0.64, close to the reported value of 0.58 for bulk PMN-PT single crystal by CTS Corporation (CTS Corporation, Bolingbrook, IL). 70 Figure 2.18. Results obtained for representative piezoelectric element 33. Left: Voltage generated by received echo and its frequency spectrum. Right: Electrical impedance magnitude and electrical impedance phase angle. 2.5.2 Electrical and Acoustical Crosstalk In order to perform crosstalk simulations, two files were created and executed: 1. Input (.flxinp) file for pulse transmission on a desired piezoelectric element and for the determination of perceived voltages on four adjacent piezoelectric elements, and 2. Review (.revinp) file for displaying results. 2.5.2a Input File for Pulse Transmission on a Desired Piezoelectric Element and for the Determination of Perceived Voltages on Four Adjacent Piezoelectric Elements This input file was essentially the same as the input file for pulse transmission of the pulse-echo response and electrical impedance up to the specification of the boundary conditions for the four sides of the array 2D model. The differences started with the specification of the driving function exciting the desired piezoelectric element for pulse transmission. In this case, the driving function consisted of a five-cycle sinusoid with an amplitude of 5 Volts at the design center frequency. The electrical window remained the same, encompassing all the 2-2 composite. The bottom electrode of the desired transmitting element was still defined as a 50 Ω series 71 resistor, whereas the bottom electrodes of the four adjacent non-transmitting electrodes were defined with receiving characteristics; each one was assigned a bottom electrode consisting of a 1 MΩ shunt resistor. Each of the five ground electrodes was defined using the ground condition and located at the top of the piezoelectric elements. Afterwards, electrical field, stress, pressure, and displacements were programmed to be calculated. The next step consisted in specifying which time histories to store in the history file. For the simulation under discussion, only bottom electrode voltages of the five piezoelectric elements were stored. Again, the last step of the input file consisted in processing the model for execution and saving all the model variables into the symbol (.symb) file for their posterior use with the review file for displaying results. 2.5.2b Review File for Displaying Results In this case, the review file was simpler since no manipulation of data was performed (e.g. no FFT was computed). It was only necessary to call up the time histories of the bottom electrode voltages belonging to the pulse transmitting element and the four adjacent non- transmitting elements to plot them as required. Figure 2.19 shows a screenshot of the results obtained for the 5-Volt, five-cycle sinusoid being applied at the design center frequency. The upper-left subfigure shows the voltage history (sec vs. V) of the transmitting piezoelectric element number 32 and the first adjacent non-transmitting piezoelectric element number 31. The upper-right subfigure shows the voltage history of transmitting element number 32 and the second adjacent non-transmitting element number 30. The lower-left subfigure shows the voltage history of transmitting element number 32 and the third adjacent non-transmitting element number 29. Finally, the lower-right subfigure shows the voltage history of transmitting element 72 number 32 and the fourth adjacent non-transmitting element number 28. Notice how the crosstalk diminishes as the adjacent non-transmitting element under consideration is farther away from the transmitting element. Figure 2.19. Results obtained for crosstalk simulation: Voltage of transmitting element vs. first adjacent non-transmitting element (upper left), vs. second adjacent non-transmitting element (upper right), vs. third adjacent non-transmitting element (lower left), and vs. fourth adjacent non-transmitting element (lower right). In order to get the crosstalk across a range of frequencies, the 5-Volt, five-cycle sinusoid was applied within a frequency range from 2 MHz to 30 MHz in steps of 2 MHz. These simulations were executed on element 32 with respective adjacent elements 31, 30, 29, and 28 as well as on element 16 with respective adjacent elements 15, 14, 13, and 12. Afterwards, the crosstalk for first adjacent elements, second adjacent elements, third adjacent elements, and fourth adjacent elements were averaged and plotted as relative amplitudes. As shown next in Figure 2.20, maximum crosstalk levels at 15 MHz were: -20.8 dB for the first adjacent element, - 23.9 dB for the second adjacent element, -24.75 dB for the third adjacent element, and -26.6 dB for the fourth adjacent element. Although these values are higher than the recommended < -35 73 dB for linear phased arrays (Table IV), one has to consider the small aperture of the linear phased array with its concomitant narrow kerfs. Should it be necessary, in order to bring the crosstalk down to < -35 dB we could include attenuating materials to the kerf filler (e.g. adding liquid plasticizer, LP-3, to the non-conductive epoxy, Epo-Tek 301), like it is done in the case of the backing layer. Figure 2.20. Computed crosstalk from 2 to 30 MHz. Maximum crosstalk levels at 15 MHz were: -20.8 dB for the first adjacent element, -23.9 dB for the second adjacent element, -24.75 dB for the third adjacent element, and -26.6 dB for the fourth adjacent element. 2.5.3 Pressure Field and Axial/Lateral Beam Profiles In order to perform these simulations, two files were created and executed: 1. Input (.flxinp) file for pressure wave generation and propagation, and 2. Review (.revinp) file for displaying the resulting pressure field and axial/lateral pressure beam profiles. 74 2.5.3a Input File for Pressure Wave Generation and Propagation The input file for pressure wave generation and propagation started with the specification of the array geometrical parameters, grid coordinates, boundary conditions, and materials exactly in the same way as the previous two sets of simulations. Additionally, while defining the coordinates of key points of the array 2D model, a routine was included to calculate the distance from each one of the piezoelectric elements to the elevation natural focal point located 8.1 mm away from the 2 nd matching layer along the dotted line that indicates the centerline of the array 2D model (Figure 2.17). For the simulation under discussion, continuous wave excitation was chosen (100-cycle sinusoid with an amplitude of 100 Vpp at the design center frequency) since it yields better approximation to theoretical lateral resolution and natural focal length. Afterwards, an electrical window was defined to encompass all the 2-2 composite; the type of transmitting electrode for the piezoelectric elements was defined once again as a 50 Ω series resistor, and was located at the bottom of each one of the piezoelectric elements (i.e. at the interface between the dielectric polyimide and the piezoelectric material). Likewise, the common ground electrode was defined and located at the top of the piezoelectric material (i.e. at the interface between the piezoelectric material and the first matching layer). Afterwards, the driving function was applied to all the piezoelectric elements, each one needing an appropriate time shift for focusing. The driving function excited the edge elements first moving towards the center, causing the emitted waves to focus at the elevation natural focal point of 8.1 mm in front of the 2 nd matching layer along the dotted line that indicates the centerline of the array 2D model. Additionally, the driving function 75 was apodized (amplitude apodization) for a smooth decay towards the edges following a half cosine wave shape. Displacements and maximum/minimum pressures were chosen to be calculated and stored into time history files. A novelty of PZFlex pressure wave generation and propagation simulations consists in the generation of a movie (.avi) file. For this purpose, graphing commands were incorporated to the routine providing real-time feedback regarding the completion percentage of the simulation. On this basis, the displayed results were updated each time the completion percentage was updated too. All the time frames were then saved and compressed for the posterior generation of the movie file. Figure 2.21 shows the output at one time frame. The top subfigure shows the pressure wave (Pa) being generated and propagating through the loading medium. The bottom subfigure shows the pressure history (sec vs. Pa) in the loading medium immediately in front of the center of the array transducer. Figure 2.21. Time frame screenshot: Pressure wave being generated and propagating through the loading medium (top) and pressure history in the loading medium immediately in front of the center of the array transducer (bottom). 76 Finally, before ending the input file for pressure wave generation and propagation, all the array transducer model variables were saved into the symbol file for their posterior use with the review file for displaying results. 2.5.3b Review File for Displaying the Resulting Pressure Field and Axial/Lateral Beam Profiles The review file for displaying the resulting pressure field and axial/lateral beam profiles started by making sure that the required input file had previously been run. The next section consisted in reading all the data (variables and time histories) generated from running the input file. In this case, the review file was also simple since no manipulation of data was performed (e.g. no FFT was computed). It was only necessary to call up the time histories of pressure maxima in the loading medium to plot them as required. Figure 2.22 shows a PZFlex screenshot of the resulting pressure field in the loading medium (Pa); notice how the pressure reaches its maximum at the elevation natural focal point of 8.1 mm. For the axial/lateral beam profiles we opted for saving and exporting the results as .dat files in order to plot them with MATLAB. Figure 2.23.A shows the axial beam profile in the loading medium along the center line of the array 2D model (mm vs. dB); notice how the axial intensity (relative pressure) oscillates in the near field, reaches its maximum at the elevation natural focal point of 8.1 mm (transition point), and decays in the far field. Figure 2.23.B shows the lateral beam profile in the loading medium across the elevation natural focal plane (mm vs. dB), with a side-lobe level (SLL) of -26 dB, and with the -6 dB beam width yielding a lateral resolution of 450 μm. 77 Figure 2.22. Pressure field in the loading medium of the SL-ERUS linear phased array under continuous wave excitation (100-cycle sinusoid with an amplitude of 100 Vpp at the design center frequency). Total loading medium length is 16 mm. (A) (B) Figure 2.23. (A) Axial beam profile in the loading medium along the centerline of the array. (B) Lateral beam profile in the loading medium across the elevation natural focal plane at 8.1mm in front of the array. 78 Table VIII below comprehensively summarizes the SL-ERUS linear phased array design parameters and simulated performance results. Table VIII. SL-ERUS linear phased array design parameters and simulated performance. Design Parameters Design center frequency 15 MHz Number of elements 64 Composite configuration 2-2, linear Pitch (0.5 water) 50 m Element width 37 m Kerf width 13 m Elevation aperture 1.8 mm Azimuth aperture 3.2 mm Elevation focal distance (natural) 8.1 mm Piezoelectric material PMN-PT 53 m thick Kerf filler material Non-conductive epoxy (Epo-Tek 301) 53 m thick Backing material Non-conductive epoxy (Epo-Tek 301) with 43.2wt% liquid plasticizer (LP-3) 2 mm thick 1 st Matching layer material Conductive epoxy (2,3- m silver epoxy) 45 m thick 2 nd Matching layer material Vapor deposited polymer (Parylene C) 30 m thick Simulated Performance -6 dB center frequency 14.1 MHz -6 dB fractional bandwidth 56.7 % -6 dB / -20 dB pulse length 1 137 / 211 nsec -6 dB lateral resolution 2 450 μm Side-lobe level (SLL) 2 -26 dB Peak-to-peak sensitivity 1 238.9 mV Axial resolution 2 80.25 μm Electrical impedance 1 123.8 , -42.94° @ 15 MHz Electromechanical coupling coefficient 1 0.64 Electrical and acoustical crosstalk < -20.8 dB @ 15 MHz 1. For representative array element 33 2. At elevation natural focus 79 2.6 Array Transducer Fabrication Fabricating ultrasound array transducers is a challenging task especially when aperture miniaturization and catheter housing are involved. Furthermore, issues such as epoxy bond thickness and uniformity, crack-free piezoelectric crystal and electrode layers, as well as material losses play a primordial role in determining the performance of a device at high frequencies. This section describes the techniques and materials used to fabricate the SL-ERUS linear phased array. The array 2-2 composite was fabricated using mechanical dicing via the dice-and-fill method. The custom flexible circuit was used to connect the array elements to the high- capacitance coaxial cable assembly consisting in a 1.5-m long bundle of 64 individual micro- coaxial cables (48 AWG), which in turn were terminated onto a pair of micro-coax boards inside of the proprietary Verasonics Backshell Connector. The array was housed inside of a 10F catheter fabricated out of biocompatible polytetrafluoroethylene (PTFE). 2.6.1 Array Acoustic Stack (2-2 Composite + 1 st Matching Layer) The array PMN-PT 2-2 composite was fabricated by CTS Corporation (CTS Corporation, Bolingbrook, IL) using the dice-and-fill technique (Savakas et al., 1981), whereby kerfs were mechanically cut with a dicing saw (0.5 mm/sec feed rate and 40,000 RPM spindle speed) into a plate of bulk PMN-PT and backfilled with non-conductive Epo-Tek 301 epoxy (Epoxy Technology, Inc., Billerica, MA). Research/industrial collaborative studies in France have reportedly used this technique to successfully build arrays of up to 20 MHz center frequency (Lethiecq et al., 1994). 80 The diced-and-filled PMN-PT plate had one of its sides then lapped down until exposing the piezoelectric elements and achieving a surface as flat as possible. 2,000-grit sandpaper together with 9-μm aluminum oxide (Al₂O₃) powder (Buehler-Illinois Tool Works, Inc., Lake Bluff, IL) were used to ensure a smooth, matte finish. Additionally, lapping with 9-μm diamond suspension (MetaDi) (Buehler-Illinois Tool Works, Inc., Lake Bluff, IL) was done at the end to remove any excess of kerf filler epoxy and make sure that the whole first lapped side of the piece was completely flat. Afterwards, the sample was mechanically cleaned by applying acetone, reagent alcohol, and Alconox detergent (Alconox, Inc., White Plains, NY) with a cotton swab and any residues left were removed with deionized water. The lapped and mechanically cleaned side of the 2-2 composite was then plasma cleaned (argon 25 sccm, 30 watts, 185 sec) to activate the exposed surface of the piezoelectric elements for electroplating sputtering of their active electrodes having a thickness of 500 Å of chrome and 2,000 Å of gold. Since electroplating sputtering actually covered the whole top surface of the 2-2 composite, the active electrodes were patterned on the piezoelectric elements by removing gold and chrome from the epoxy kerfs with a cotton swab dampened in reagent alcohol, taking advantage of the weak adhesion of chrome and gold to epoxies. The 2-2 composite was then flipped over and lapped down to its final design thickness. Again, 2,000-grit sandpaper together with 9-μm aluminum oxide (Al₂O₃) powder were used to ensure a smooth, matte finish and 9-μm diamond suspension lapping was done to remove any excess of kerf filler epoxy. The exposed second side of the 2-2 composite underwent the same process of mechanical and plasma cleaning, together with electroplating sputtering now of the common ground electrode, reason why there was no electrode patterning on this occasion. 81 The 1 st matching layer was then prepared by adding 4.5 gr of 2,3- m silver powder (Aldrich Chemicals Co., St. Louis, MO) to 1.25 gr of mixed of non-conductive epoxies: Insulcast 501 and Insulcure 9 (ITW Polymers Coatings North America, Montgomeriville, PA). The resulting mixture was then degassed to avoid the presence of air bubbles, cast on top of the common ground electrode, and centrifuged at 3,000 RPM for 15 mins to distribute it evenly. After curing overnight in a dry-environment nitrogen box and post-curing in a conventional oven for 2 hrs at 45 °C, the 1 st matching layer was lapped down to its final design thickness and the resulting acoustic stack (2-2 composite + 1 st matching layer) was mechanically diced (0.5 mm/sec feed rate and 30,000 RPM spindle speed) to final dimensions (Figure 2.24). Figure 2.24. Acoustic stack comprising the PMN-PT 2-2 composite and 1 st matching layer. Active and ground electrodes were sputtered on the bottom and top sides of the composite, respectively. Both layers were lapped down to final design thicknesses and the resulting stack was mechanically diced to final dimensions. 2.6.2 Flexible Printed Circuit The flexible circuit was fabricated by MicroConnex (MicroConnex Corp., Snoqualmie, WA). Figure 2.25 shows a physical sample of the flexible circuit consisting of a 25-μm-thick base dielectric polyimide layer (Kapton) (DuPont, Wilmington, DE), which features cover-layer protected signal traces on the front (left subfigure) and an exposed common ground electrode on 82 the back with windows for 2-2 composite proper alignment and positioning (center subfigure). All signal traces and ground electrodes were patterned over a 4-μm-thick copper layer and finished with a flash of gold. Before continuing the fabrication process, a fitting test was performed by folding the flexible printed circuit into its final configuration and placing it inside of a piece of 10F polyimide tubing (right subfigure). Figure 2.25. Flexible printed circuit for the SL-ERUS linear phased array. Left: Front side with cover-layer protected signal traces. Center: Back side with exposed common ground electrode featuring three windows for 2-2 composite proper alignment and positioning. Right: Fitting test of flexible circuit folded into its final configuration and placed inside of a 10F polyimide tube. 2.6.3 Bonding the Array Acoustic Stack and Backing Layer to the Flexible Printed Circuit After mechanical dicing to final dimensions, both sides of the acoustic stack were cleaned by applying trichloroethylene, acetone, and reagent alcohol with a cotton swab. Likewise, the front side of the flexible circuit was cleaned with a cotton swab dampened in reagent alcohol. Afterwards, a thin line of Epo-Tek 301 epoxy was applied with a foam swab at the center of the flexible circuit (2-2 composite area). Another foam swab was used then to smear a thin layer of Epo-Tek 301 epoxy all over the surface of the 2-2 composite (bottom side of acoustic stack). As shown in Figure 2.26, the acoustic stack was afterwards placed on top of the flexible circuit (left subfigure) visually making sure that the piezoelectric elements were correctly aligned with the signal traces (center subfigure), and pressure was applied using a “C” clamp (right subfigure). 83 The resulting sub-assembly was cured overnight in a dry-environment nitrogen box and then post-cured in a conventional oven for 2 hrs at 45 °C. Figure 2.26. Bonding the array acoustic stack to the flexible printed circuit. Left: Array acoustic stack placed on top of the flexible circuit. Center: Piezoelectric elements correctly aligned with the signal traces of the flexible circuit. Right: Pressure applied with a “C” clamp to bond the resulting sub-assembly. The next step consisted in bonding the backing layer to the backside of the flexible circuit. For this purpose, the backside of the flexible circuit was cleaned with a cotton swab dampened in reagent alcohol. Afterwards, a thin line of Epo-Tek 301 epoxy was applied with a foam swab at the center of the flexible circuit (backside of 2-2 composite area) and another foam swab was used then to smear a thin layer of Epo-Tek 301 epoxy all over the bonding surface of the backing layer. The backing layer was afterwards placed on top of the flexible circuit backside and pressure was once again applied with a “C” clamp. Likewise, the resulting sub-assembly was cured overnight in a dry-environment nitrogen box and then post-cured in a conventional oven for 2 hrs at 45°C. Figure 2.27.A shows a CAD rendered image of the resulting sub-assembly with the array acoustic stack and backing layer bonded to the flexible circuit. In order to connect the common ground electrode on the backside of the flexible circuit to the front matching layer, the two grounding flaps featuring metallized thru-vias were bent (Figure 2.27.B) and then 84 bonded to the front matching layer using conductive epoxy, E-Solder 3022 (Von Roll Isola USA, Inc., Schenectady, NY), (Figure 2.27.C). Afterwards, the two main branches of the flexible circuit were bent and bonded to the sides of the backing layer using Epo-Tek 301 epoxy. Figure 2.27.D shows a picture of the resulting sub-assembly, which was then all covered (excluding the soldering pads) with vapor deposited polymer, Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN), serving as both the 2 nd matching layer and insulation. Figure 2.28 shows cross- sectional images revealing the diverse components of the sub-assembly and a detail view of the bending of one main branch of the flexible circuit, making sure there were no broken signal traces. Figure 2.27. (A) CAD rendered image of the array acoustic stack and backing layer bonded to the flexible circuit. (B) Bending the two grounding flaps in order to connect the ground layer on the backside of the flexible circuit to the front matching layer through metallized thru-vias. (C) Grounding flaps bonded to the front matching layer with conductive E-Solder 3022 epoxy. (D) Resulting sub-assembly with two main branches of the flexible circuit bonded to the sides of the backing layer with non-conductive Epo-Tek 301 epoxy. 85 Figure 2.28. (A) thru (C) Cross-sectional images revealing the diverse components of the array acoustic stack and backing layer bonded to the flexible circuit. (D) Detailed view of the bending of one main branch of the flexible circuit, making sure there were no broken signal traces. 2.6.4 Cable Assembly Connection to the Flexible Printed Circuit This section was possibly the most challenging one of the SL-ERUS probe fabrication process mainly due to the fragility of the miniature 48 AWG micro-coaxial cables (e.g. 171-μm- diameter jacket and 39-μm-diameter inner conductor), as well as the required tight soldering 86 pitch of only 180 μm. Two different approaches were attempted to achieve a working electrical interconnect solution: 1. Manual micro-soldering with solder and soldering iron, and 2. Manual cold-soldering with conductive epoxy, E-Solder 3022. 2.6.4a Manual Micro-Soldering with Solder and Soldering Iron Manual micro-soldering was attempted by Hitachi Metals (Hitachi Metals, Ltd., Tokyo, Japan). As it was expected, challenges arose when soldering the bundle of sixty-four 48 AWG micro-coaxial cables especially due to the tight pitch of only 180 μm between consecutive signal pads on the flexible circuit. First trials showed broken coaxial cores, electrical shorts between consecutive signals, and signal-to-ground electrical shorts due to melted insulation sections (Figure 2.29.A). Nonetheless, our collaborating group at Japan eventually showed that 180-μm pitch soldering by hand is possible (Figure 2.29.B) with not a single broken coaxial core and 59 out of 64 soldered cores showing no electrical shorts. Additionally, catheter-fitting tests for final packaging showed that the flexible circuit successfully fitted inside of a 10F catheter with its branches folded into final configuration and with all the sixty-four 48 AWG micro-coaxial cables already soldered (Figure 2.29.C). However, these trials were all done using 5-cm long pieces of 48 AWG micro-coaxial cable. The challenge of assembling a 1.5-m long bundle of sixty-four 48 AWG micro-coaxial cables and successfully pulling it inside of a 10F catheter without any damage required more technological and human resources that resulted in Hitachi Metals having to stop and forgo this collaboration endeavor. 87 Figure 2.29. Manual micro-soldering of sixty-four 48 AWG micro-coaxial cables to flexible printed circuit. (A) First trials showed broken coaxial cores and electrical shorts between consecutive signals. (B) 180-μm pitch soldering by hand is possible with no broken coaxial cores and 59 out of 64 soldered cores without electrical shorts. (C) Flexible circuit successfully fitted inside of a 10F catheter with its soldering branches folded into final configuration and with all the sixty-four 48 AWG micro-coaxial cables already soldered. 2.6.4b Manual Cold-Soldering with Conductive Epoxy, E-Solder 3022 Manual cold-soldering with conductive epoxy, E-Solder 3022, was attempted by our group. For this approach, the first step consisted in assembling the 1.5-m long bundle of 64 pieces of 48 AWG micro-coaxial cable and enclosing it inside of a 10F catheter considering Hitachi Metals’ previous drawbacks. For this purpose, the micro-coaxial cable was cut into sixty- four 1.5-m long pieces with their ends subsequently codified with colors and a corresponding number, respectively. Afterwards, all the 64 pieces were carefully aligned on the color end and glued together with Loctite 430 (Henkel AG & Co., Düsseldorf, Germany) inside of a 1-cm long 88 piece of 8F polytetrafluoroethylene (PTFE) tubing to which a 2-m long 1F guide wire was hooked into (Figure 2.30.A). All the 64 pieces of micro-coaxial cable were then manually pulled through a 1.45-m long piece of 10F PTFE tubing (Figure 2.30.B), and upon exiting, the 1-cm long piece of 8F PTFE tubing was cut off in order to immediately proceed to number the other end of each micro-coaxial cable according to their color code (Figure 2.30.C). (A) (B) (C) Figure 2.30. Cable assembly and packaging process. (A) All sixty-four 1.5-m long pieces of 48 AWG micro-coaxial cable carefully aligned and glued together inside of a 1-cm long piece of 8F PTFE tubing to which a 1F guide wire was hooked into. (B) All 64 pieces of micro-coaxial cable pulled through a 1.45-m long piece of 10F PTFE tubing. (C) Piece of 8F PTFE tubing cut off and other end of each micro-coaxial cable numbered. 89 The next step consisted in manually stripping each piece of 48 AWG micro-coaxial cable according to the specifications shown in Figure 2.31; a process that proved to be laborious and intensive since this fine-gauge micro-coaxial cable can become easily damaged (e.g. bent inner conductor, broken insulation, or broken inner conductor). Automatic, industrial-grade stripping services were sought (Laser Wire Solutions Ltd., Pontypridd, UK) but the scarcity of service providers for this fine gauge as well as the high lead time and costs involved (e.g. tooling and fixture development, process setup and refinement, and man-hours) resulted in our group having to forgo such an option. Figure 2.31. Stripping specifications of 48 AWG micro-coaxial cable. Afterwards, each single piece of 48 AWG micro-coaxial cable was carefully placed, aligned, and secured onto the flexible printed circuit. Once secured, conductive E-Solder 3022 epoxy was manually applied to cold-solder each 39-μm-diameter inner conductor to its 90 respective 100-μm-wide connection pad; a process that resulted to be lengthy and challenging, as only three inner conductors (one inner conductor per flexible circuit branch) could be cold- soldered per day and extreme caution had to be taken in order to avoid creating electrical shorts between consecutive connection pads having a tight pitch of only 180 μm. After completing the first side with the first 32 connections, non-conductive Epo-Tek 301 epoxy was manually applied all over the three flexible circuit branches to pot and protect the connections created (Figure 2.32.A). More E-Solder 3022 was then manually applied to cold-solder the exposed outer conductors to the ground pads of each of the three flexible circuit branches (Figure 2.32.B). The same process was repeated for the second side with the remaining 32 connections (Figure 2.32.C). Figure 2.32. Connection of the 48 AWG micro-coaxial cable assembly to the flexible printed circuit. (A) Cores aligned and cold-soldered with E-Solder 3022 to the connection pads. All created connections were potted and protected with Epo-Tek 301. (B) Exposed outer conductors cold-soldered with E-Solder 3022 to the ground pads. (C) All 64 connections completed. 91 2.7 Final Packaging and Connection to Verasonics Backshell Connector After cold-soldering the cable assembly, the soldering branches of the flexible printed circuit were folded into final configuration (Figure 2.33.A) and carefully pulled inside the distal end of the 10F PTFE catheter (Figure 2.33.B). The array aperture was then positioned right in the center of a manually pre-cut window (3.7 mm x 2.2 mm) on the distal end of the catheter. Epo- Tek 301 was applied to fill the gaps between the acoustic stack-flexible circuit assembly and the edges of the window, as well as to seal the opening at the tip of the catheter. Afterwards, a 7-μm layer of Parylene C was vapor deposited to ensure complete insulation of the distal end of the SL-ERUS phased array probe (Figure 2.33.C). A size comparison of the distal end of the SL- ERUS phased array probe, the array acoustic stack–backing layer–flexible circuit sub-assembly, and a dime coin is shown in Figure 2.33.D. Figure 2.33. Final packaging of the SL-ERUS phased array probe. (A) Soldering branches of the flexible printed circuit folded into final configuration. (B) Assembly pulled inside of the 10F PTFE catheter. (C) Finished distal end with 7-μm layer of vapor deposited Parylene C to ensure complete insulation. (D) Size comparison of the distal end of the SL-ERUS phased array probe, the array acoustic stack–backing layer–flexible circuit sub-assembly, and a dime coin. 92 The proximal end of the cable assembly was then manually cold-soldered with E-Solder 3022 to a pair of Verasonics micro-coax termination boards (Figure 2.34.A). All the connections were potted with Epo-Tek 301 for protection, leaving exposed just the top surface of the resulting E-Solder 3022 electrodes for posterior poling and testing. The two bundles of 32 micro- coaxial cables were each held together with a piece of polyimide tape (Kapton) (DuPont, Wilmington, DE) (Figure 2.34.B), protected with a piece of electrical shielding tape (3M Company, Maplewood, MN), bonded with Epo-Tek 301 to a black PVC cable-strain relief boot (Amphenol Corporation, Wallingford, CT) (Figure 2.34.C), and finished with a Verasonics M14- A2-70 Hex cable-strain relief nut (Figure 2.34.D). Figure 2.34. Cable assembly termination of the proximal end. (A) 48 AWG micro-coaxial cables cold-soldered to Verasonics termination boards. (B) Two bundles of 32 micro-coaxial cables each held together with a piece of Kapton tape. (C) Protection with electrical shielding tape and black PVC cable-strain relief boot. (D) Finish with a Verasonics M14-A2-70 Hex cable-strain relief nut. 93 To connect the resulting SL-ERUS phased array probe (Figure 2.35.A) to the Verasonics Backshell Connector, termination board 1L (channels 1 to 32) was connected to the Left Canister PCB (printed circuit board) and termination board 1R (channels 33 to 64) was connected to the Right Canister PCB (Figure 2.35.B). Afterwards, the connected set of termination boards and PCBs together with the Cannon DL-260 connector plug and locking handle were fitted inside one backshell half (Figure 2.35.C) and finally the other backshell half was secured in place with the included set of screws to completely finish the SL-ERUS phased array probe (Figure 2.35.D). (A) 94 (B) (C) (D) Figure 2.35. Connecting the SL-ERUS phased array probe to the Verasonics Backshell Connector. (A) Probe finished with Verasonics micro-coax termination boards. (B) Connection schematics of the termination boards to the left and right canister printed circuit boards. (C) Fitting Verasonics connection electronics inside one backshell half. (D) Completely finished and packaged SL-ERUS phased array probe. 95 2.8 Array Transducer Characterization Upon fabrication completion of the SL-ERUS phased array probe, the next step consisted in individually re-polarizing each one of the 64 piezoelectric PMN-PT elements in air at room temperature under an electric field of 20 kV/cm for 3 minutes using a Spellman-Bertan Model 210-02R high voltage power supply (Spellman High Voltage Electronics Corporation, Hauppauge, NY). Afterwards, several standard non-imaging transducer tests were performed on the array probe to characterize its performance including electrical impedance, pulse-echo response, insertion loss, combined electrical and acoustical crosstalk, as well as single-element azimuthal one-way angular response or directivity. 2.8.1 Electrical Impedance The electrical impedance of each individual array element was measured with an Agilent E4991A RF Impedance/Material Analyzer (Agilent Technologies, Santa Clara, CA) and both magnitude and phase angle were recorded over the frequency range of the array transducer pass- band. Since the measurements were made on the Verasonics micro-coax termination board electrodes, with the micro-coaxial cables already connected to their respective array elements through the flexible printed circuit, it was necessary to determine the electrical impedance of each load (i.e. each array element) out from the measured input electrical impedance value of each transmission line. The input electrical impedance (Zin) of a transmission line is defined as (Pozar, 2012): 𝑍 𝑖𝑛 = 𝑍 0 ( 𝑍 𝐿 +𝑗 𝑍 0 tan (𝛽𝑙 ) 𝑍 0 +𝑗 𝑍 𝐿 tan (𝛽𝑙 ) ) (2.2) 96 where Z0 is the characteristic impedance, ZL is the load impedance, β is the phase constant, and l is the length of the transmission line. Additionally: 𝛽 = 2𝜋 𝜆 (2.3) 𝜆 = 𝜆 0 √ 𝜀 𝑟 (2.4) 𝜆 0 = 𝑐 𝑓 (2.5) where λ is the electromagnetic wavelength in the transmission line, λ0 is the electromagnetic wavelength in free space, εr is the relative permittivity of the transmission line dielectric layer, c is the speed of light in free space, and f is the frequency of interest. Solving equation (2.2) for ZL yields: 𝑍 𝐿 = 𝑍 𝑖𝑛 −𝑗 𝑍 0 tan (𝛽𝑙 ) 1−𝑗 𝑍 𝑖𝑛 𝑍 0 tan (𝛽𝑙 ) (2.6) For each 48 AWG micro-coaxial cable used: Z0 = 50 Ω, l = 1.5 m, εr = 2.1 (Perfluoroalkoxy – PFA), c = 3 x 10 8 m/s, and f = 1 MHz to 30 MHz. The simulated and measured electrical impedance magnitude and phase angle of a representative array element (element 33) are shown in Figures 2.36.A and 2.36.B, respectively. Element 33 was chosen as representative since its electrical impedance characteristics were closest to the average of all 64 elements. 97 (A) (B) Figure 2.36. (A) Simulated and (B) measured electrical impedance magnitude and phase angle of representative array element 33. The simulated and measured results are summarized and compared in Table IX. The measured results showed an electrical impedance magnitude of 131.3 ± 27.1 Ω at 15 MHz. The series (fr) and parallel (fa) resonant frequencies were 17.6 ± 1.2 MHz and 20.7 ± 1.6 MHz, respectively, yielding an electromechanical coupling coefficient (kt) of 0.57 ± 0.05, close to the reported value of 0.58 for bulk PMN-PT single crystal by CTS Corporation (CTS Corporation, Bolingbrook, IL). Table IX. Comparison of simulated and measured electrical impedance results. Parameter PZFlex Measured Electrical Impedance ( ) 123.8 @ 15 MHz 131.3 ± 27.1 @ 15 MHz fr (MHz) 15.8 17.6 ± 1.2 fa (MHz) 19.9 20.7 ± 1.6 kt 0.64 0.57 ± 0.05 The uniformity of the measured values of electrical impedance magnitude and phase angle of all the array elements at 15 MHz is shown in Figure 2.37. As it can be noticed, there were only four open elements: 42, 54, 55, and 64. Previously, these four elements presented a short circuit to ground, reason why the connections were opened by cutting the inner conductors cold-soldered to their respective electrodes on the Verasonics micro-coax termination board. The 98 average and standard deviation of the array elements electrical impedance magnitude and phase angle were (131.3 ± 27.1 Ω) and (-35.2 ± 5.4°), respectively. Figure 2.37. Uniformity of electrical impedance magnitude and phase angle values for all the array elements. 2.8.2 Pulse-Echo Response and Insertion Loss The pulse-echo response of each individual array element was recorded to determine their effective center frequency, bandwidth, sensitivity, and pulse length. This test was performed by immersing the SL-ERUS phased array probe in a deionized water tank containing a polished quartz reflector as the target at the elevation natural focal distance of 8.1 mm. The pulser/receiver used was a Panametrics – NDT 5900PR (Panametrics, Inc., Waltham, MA), which emitted single cycle unipolar pulses with an amplitude of -100 V at a pulse repetition frequency (PRF) of 200 Hz. Energy and damping were set at 1 J and 50 Ω, respectively. To receive echo signals, the Panametrics – NDT 5900PR was set with a bandpass filter from 10 99 MHz to 20 MHz and a gain of 40 dB, which were applied before analog signals were digitized via a GaGe EON CompuScope CS122G1 Digitizer (Dynamic Signals, LLC, Lockport, IL) with a 1-GS/s sampling rate. For each pulse-echo time domain signal, the -6 dB fractional bandwidth was determined using a Fast Fourier Transform (FFT), where the lower and upper bandwidth edges were determined by the frequencies at which the power spectrum was equal to -6 dB relative to the maximum value. The effective center frequency was taken as the middle point between the lower and upper limits of the -6 dB fractional bandwidth. Echo peak-to-peak amplitude was recorded for sensitivity and the -6 dB / -20 dB signal pulse lengths were determined by measuring the time between the first and last points, at which the signal was -6 dB / -20 dB relative to the maximum echo signal value, respectively. Insertion loss was measured by immersing the SL-ERUS phased array probe in a deionized water tank, exciting each element in the array with a 5-Vpp, 30-cycle sinusoidal tone- burst signal generated by a Tektronix AFG 3252 Dual Channel Arbitrary / Function Generator (Tektronix, Inc., Beaverton, OR) at the array center frequency, and receiving the reflected echo from a polished quartz reflector placed at the elevation natural focal distance. The receive power across a 50-Ω load was referenced to the source power delivered to a 50-Ω reference load and expressed in decibels (Cannata et al., 2005). The measured values were afterwards corrected for loss due to attenuation in water (2.0 x 10 -4 dB/mm-MHz 2 ) (Lockwood et al., 1994) and reflection from the quartz target (1.9 dB). The simulated and measured pulse-echo response of a representative array element (element 37) are shown in Figures 2.38.A and 2.38.B, respectively. Element 37 was chosen as 100 representative since its pulse-echo response characteristics were closest to the average of all 64 elements. (A) (B) Figure 2.38. (A) Simulated and (B) measured pulse-echo response of representative array element 37. The simulated and measured results are summarized and compared in Table X. The measured results showed an effective -6 dB center frequency (-6 dB Fc) of 17.7 ± 1.2 MHz, a -6 dB fractional bandwidth (-6 dB BW) of 52.2 ± 9.8%, a peak-to-peak sensitivity (Vpp) of 200.1 ± 102.7 mV, and a compensated insertion loss (IL) of 49.8 ± 3.9 dB. The increase of center frequency resulted from slightly over lapping the 2-2 composite by hand, which consequently resulted in shorter -6 dB / -20 dB pulse lengths. The overall decrease in sensitivity resulted mainly from the fact that not all the array elements had their entire bottom surface perfectly connected to their respective flexible circuit signal electrodes, as it was the case of the PZFlex simulation model. In reality, the resulting Epo-Tek 301 epoxy thin bonding layer was not perfectly uniform after the clamping process. This can be verified by closely examining the cross-sectional image of the array acoustic stack shown in Figure 2.28.B and by observing the peak-to-peak sensitivity uniformity (Figure 2.39), which clearly shows how the second half of the array (elements 33 to 64) was better connected to the flexible printed circuit by showing 101 higher peak-to-peak sensitivities than the first 32 elements. Furthermore, another cause for the decrease in sensitivity was the additional 7-μm layer of Parylene C that was vapor deposited to ensure complete insulation of the distal end of the SL-ERUS phased array probe. The measured compensated insertion loss was higher than that reported for other linear phased arrays with similar characteristics; however, the element footprint and, consequently, the dimensions of the signal electrodes on the flexible printed circuit in this array were at least 2 times smaller, making it more difficult to ensure a perfect connection between the bottom surface of the array elements and their respective flexible circuit signal electrodes. For example, the side- looking phased array developed by Stephens et al. (Stephens et al., 2008) featured 64 elements within an aperture of 2.6 mm by 6.4 mm, at a pitch of 100 μm, and a center frequency of 7.25 MHz; whereas the side-looking phased array being presented included the same number of elements within an aperture of 1.8 mm by 3.2 mm, at a pitch of only 50 μm, and a center frequency of 17.7 MHz. Additionally, the 64 individual 48-AWG micro-coaxial cables used in the array under discussion were not shielded as a bundle, making them more susceptible to external noise and interference. A solution could be to assemble them together with 50-μm binder tape, shield them with 46-AWG braiding, and wrap them with an additional 50-μm fluoropolymer tape (Stephens et al., 2008). This could not only decrease insertion loss but also improve sensitivity. Table X. Comparison of simulated and measured pulse-echo response results. Parameter PZFlex Measured -6 dB Fc (MHz) 14.1 ± 0.5 17.7 ± 1.2 -6 dB BW (%) 56.7 ± 2.1 52.2 ± 9.8 Vpp (mV) 238.9 ± 98.7 200.1 ± 102.7 -6 dB / -20 dB pulse length (nsec) 137 / 211 107 / 193 Compensated IL (dB) N/A 49.8 ± 3.9 102 The uniformity of the measured values of -6 dB center frequency, -6 dB fractional bandwidth, and peak-to-peak sensitivity of all the array elements is shown in Figure 2.39. Figure 2.39. Uniformity of -6 dB center frequency, -6 dB fractional bandwidth, and peak-to- peak sensitivity of all the array elements. 2.8.3 Electrical and Acoustical Crosstalk Combined electrical and acoustical crosstalk measurements were performed by immersing the SL-ERUS phased array probe in a deionized water tank. A Tektronix AFG 3252 Dual Channel Arbitrary / Function Generator (Tektronix, Inc., Beaverton, OR) was used to excite one element in the array with a 5-Vpp, 5-cycle sinusoidal tone-burst signal within a frequency range from 2 MHz to 30 MHz in steps of 2 MHz. The applied signal was measured with a LeCroy LC534 1-GHz Oscilloscope (LeCroy Corporation, Chestnut Ridge, NY) and served as a reference to the measured signals from four adjacent elements. These measurements 103 were executed on element 18 with respective adjacent elements 17, 16, 15, and 14 as well as on element 37 with respective adjacent elements 36, 35, 34, and 33. Afterwards, the crosstalk for first adjacent elements, second adjacent elements, third adjacent elements, and fourth adjacent elements were averaged and plotted as relative amplitudes. The simulated and measured crosstalk values are shown in Figures 2.40.A and 2.40.B, respectively. (A) (B) Figure 2.40. (A) Simulated and (B) measured crosstalk values for four nearest neighboring elements in the array. Maximum measured crosstalk values at the array center frequency were -30.5 dB for the first adjacent element, -32.8 dB for the second adjacent element, -34.4 dB for the third adjacent element, and -40.1 dB for the fourth adjacent element; which were in accordance to the suggested < -30 dB crosstalk design guideline for arrays with a linear configuration (Ritter et al., 2002). As it can be noticed, all the measured values of crosstalk were lower than their simulated counterparts. This mainly resulted from the fact that there were minuscule void spaces (cross- sectional area of 20 μm x 4 μm) defined by adjacent signal electrodes on the flexible printed circuit and the 2-2 composite, whereas in the simulation model such void spaces were not taken into consideration. 104 2.8.4 Single-Element Directivity Response The one-way azimuthal directivity response was measured by rotating a representative array element around an axis along its length and center. The array element was excited with a 50-Vpp, 5-cycle sinusoidal tone-burst signal generated by a Tektronix AFG 3252 Dual Channel Arbitrary / Function Generator connected to an Amplifier Research 50W1000B solid-state power amplifier (Amplifier Research, Inc., Souderton, PA). A needle hydrophone HGL-0085 (Onda Corporation, Sunnyvale, CA), placed at the natural elevation focus and connected to a LeCroy WaveRunner 104MXi 1-GHz Oscilloscope, was used to acquire the amplitude of the time- domain response at discrete angular positions (Cannata et al., 2005). The Field II program (Jensen, 1996) was used to simulate the directivity of a single element in the array and to estimate the effective element width by matching the obtained simulated directivity curve to the measured values (Cannata et al., 2006). The simulated and measured one-way azimuthal directivity response of representative array element 37 are shown in Figure 2.41. Figure 2.41. Simulated and measured one-way azimuthal directivity response of representative array element 37. 105 The measured -6 dB directivity was approximately ±22°. The desired measured crosstalk explains why the effective element width (44 μm) was only 19% larger than the actual element width of 37 μm. A similar result was reported by Ritter et al. of an effective element width 16% larger than the actual element width for a 30-MHz linear array with desired maximum crosstalk < -30 dB and -6 dB directivity of ±20° (Ritter et al., 2002). Likewise, Chiu et al. reported a -6 dB directivity of approximately ±20° for a 20-MHz phased array with close-to-desired maximum crosstalk < -28 dB (Chiu et al., 2017). Table XI comprehensively summarizes and compares the simulated and measured results for the array transducer characterization testing. Table XI. Comparison summary of simulated and measured characterization results. Parameter PZFlex Measured Electrical impedance ( ) 123.8 @ 15 MHz 131.3 ± 27.1 @ 15 MHz Series resonant frequency, fr (MHz) 15.8 17.6 ± 1.2 Parallel resonant frequency, fa (MHz) 19.9 20.7 ± 1.6 Electromechanical coupling coefficient, kt 0.64 0.57 ± 0.05 -6 dB center frequency, Fc (MHz) 14.1 ± 0.5 17.7 ± 1.2 -6 dB fractional bandwidth (%) 56.7 ± 2.1 52.2 ± 9.8 Peak-to-peak sensitivity (mV) 238.9 ± 98.7 200.1 ± 102.7 -6 dB / -20 dB pulse length (nsec) 137 / 211 107 / 193 Compensated insertion loss (dB) N/A 49.8 ± 3.9 Electrical and acoustical crosstalk (dB) < -20.8 < -30.5 Single-element directivity response 1 ± 22° ± 22° 1. Simulated value obtained with Field II program. 2.9 Array Transducer Imaging The ultimate performance indicators of the SL-ERUS phased array probe were determined by its imaging capability of four different kinds of targets using a Verasonics Vantage 128 System. Imaging was performed using multiple ray lines in a phased array 106 configuration, with each ray line constituting a separate beamformed transmit operation, followed by a respective beamformed receive operation. Radio frequency (RF) data was acquired using the Verasonics Nyquist sampling mode (Verasonics, 2017), with the transmit frequency set to 15.625 MHz. After all of the individual ray lines were acquired, the image data was reconstructed to yield the phased array images. No apodization was applied to the array transducer aperture in any of the imaging experiments. 2.9.1 Polished Quartz Reflector Imaging The first target (Figure 2.42) was a polished quartz reflector immersed in a tank with deionized water and placed at three different depths: 1. Half the elevation natural focal distance, 2. Elevation natural focal distance, and 3. Twice the elevation natural focal distance. For these three cases, respectively, the transmit focus was set at a depth of 4 mm with a transmit pulse of 18 Vpp, a depth of 8 mm with a transmit pulse of 40 Vpp, and a depth of 16 mm with a transmit pulse of 16 Vpp. The acquired images were post-processed with a scaling gain of 40 dB. Figure 2.42. Polished quartz reflector for imaging experiments of the SL-ERUS phased array probe. The polished quartz reflector was immersed in a tank with deionized water. 107 The acquired images of the polished quartz reflector are shown in Figure 2.43. As it can be seen, the array probe was capable of imaging such target up to a depth of twice the elevation natural focal distance. As expected, when the quartz reflector was placed in the near field at half the elevation natural focal distance (Figure 2.43.A) the field of view was reduced and cluttering could be seen all along the imaged surfaces. Both, top and bottom surfaces were visible and a second reflection from the top surface could also be observed. When the quartz reflector was placed at the elevation natural focal distance (Figure 2.43.B) the field of view increased and cluttering was reduced in the central region of the imaged surfaces. Again, both top and bottom surfaces were clearly visible. When the quartz reflector was placed in the far field at twice the elevation natural focal distance (Figure 2.43.C) the field of view continued to increase and cluttering remain reduced in the central region of the imaged surfaces. Both, top and bottom surfaces were clearly visible once again. (A) (B) (C) Figure 2.43. Imaging of polished quartz reflector. (A) Target located in the near field at half the elevation natural focal distance. (B) Target located at the elevation natural focal distance. (C) Target located in the far field at twice the elevation natural focal distance. 108 2.9.2 Custom-Made Fine-Wire Phantom Imaging The second target (Figure 2.44) was a custom-made fine-wire phantom composed of three evenly spaced 20-μm-diameter tungsten wires (California Fine Wire Company, Grover Beach, CA) with the finality of determining the axial and lateral spatial resolutions of the array transducer. The axial separation between wires was around 1.0 mm and the azimuth separation was around 1.6 mm. The custom-made fine-wire phantom was immersed in a tank with deionized water. At a center frequency of 15 MHz, the sound wavelength in the transmitting medium is 100 μm; therefore, the wires could safely be assumed as three point targets. The transmit focus was set at a depth of 8 mm with a transmit pulse of 40 Vpp. The acquired image was post-processed with a scaling gain of 15 dB and minimal thresholding was used in order to highlight the wire targets. Figure 2.44. Custom-made fine-wire phantom for imaging experiments of the SL-ERUS phased array probe. The phantom was immersed in a tank with deionized water. The acquired image of the custom-made fine-wire phantom is shown in Figure 2.45.A. As it can be seen, the three wires were clearly visible with reasonable image quality, especially the middle one located at the elevation natural focal distance. Faint artifacts were observed 109 mostly to the left and right sides of the three wires due to the presence of sidelobes in the ultrasound beam. Implementing apodization could have reduced the amplitude of such sidelobes; however, this would have increased the main lobe width, consequently degrading spatial resolutions (Frazier and O’Brien, 1998). Faint artifacts were also observed on the front of the wires mainly due to thickness irregularities of the matching layers as well as slight wire reverberation and misalignment with respect to the array transducer elevation aperture. Plots of the axial and lateral line spread functions for the middle wire are shown in Figures 2.45.B and 2.45.C, respectively. The measured full-width half-maximum (FWHM) spatial resolutions were around 90 μm and 420 μm in the axial and lateral directions, respectively. These measurements correlated well with the theoretical axial (80.25 μm) and lateral (450 μm) spatial resolutions, predicted by equations (1.20) to (1.22). Figure 2.45. Imaging of custom-made fine-wire phantom. (A) Acquired image. (B) Plot of the axial line spread function for the middle wire. (C) Plot of the lateral line spread function for the middle wire. 110 2.9.3 Custom-Made Tissue-Mimicking Phantom with Needle Imaging The third target (Figure 2.46) was a custom-made tissue-mimicking phantom composed of urethane rubber (α = 0.5 dB/cm-MHz, c = 1,450 m/s) (Supertech, Inc., Elkhart, IN) into which a 700-μm-diameter stainless steel needle was inserted with the finality of demonstrating the capability of the array transducer to detect solid structures embedded in tissue. The custom-made tissue-mimicking phantom was immersed in a tank with deionized water. The transmit focus was set at a depth of 12 mm with a transmit pulse of 40 Vpp. The acquired images were post- processed with a scaling gain of 26 dB. Figure 2.46. Custom-made tissue-mimicking phantom with inserted 700-μm-diameter stainless steel needle for imaging experiments of the SL-ERUS phased array probe. The phantom was immersed in a tank with deionized water. The acquired images of the custom-made tissue-mimicking phantom with an inserted 700-μm-diameter stainless steel needle are shown in Figure 2.47. Figure 2.47.A shows the needle inserted at a depth of approximately 3 mm and oblique across the array transducer azimuth aperture, reason why the needle longitudinal wall is also visible. Figure 2.47.B shows the needle inserted at a depth of approximately 5 mm and oblique across the array transducer azimuth 111 aperture. Figure 2.47.C shows the needle inserted at a depth of approximately 8 mm and oblique across the array transducer azimuth aperture. Figure 2.47.D shows the needle inserted at a depth of approximately 8 mm and parallel across the array transducer elevation aperture, reason why the needle longitudinal wall is not visible in this case. Figure 2.47. Imaging of custom-made tissue-mimicking phantom with inserted stainless steel needle. (A) Needle inserted at a depth of 3 mm and oblique across the array transducer azimuth aperture. (B) Needle inserted at a depth of 5 mm and oblique across the array transducer azimuth aperture. (C) Needle inserted at a depth of 8 mm and oblique across the array transducer azimuth aperture. (D) Needle inserted at a depth of 8 mm and parallel across the array transducer elevation aperture. 2.9.4 Custom-Made Tissue-Mimicking Phantom with Tubing Imaging The fourth and last target (Figure 2.48) was a custom-made tissue-mimicking phantom composed of urethane rubber into which a piece of 3-mm-diameter polyimide tubing was inserted with the finality of demonstrating the capability of the array transducer to detect cysts 112 present in tissue. The custom-made tissue-mimicking phantom was immersed in a tank with deionized water. The transmit focus was set at a depth of 6 mm with a transmit pulse of 40 Vpp. The acquired images were post-processed with a scaling gain of 22 dB. Figure 2.48. Custom-made tissue-mimicking phantom with inserted piece of 3-mm-diameter polyimide tubing for imaging experiments of the SL-ERUS phased array probe. The phantom was immersed in a tank with deionized water. The acquired images of the custom-made tissue-mimicking phantom with an inserted piece of 3-mm-diameter polyimide tubing are shown in Figure 2.49. Figure 2.49.A shows the cyst resulting from inserting the piece of polyimide tubing at a depth of approximately 5 mm and parallel across the array transducer elevation aperture. Similarly, Figure 2.49.B shows the cyst resulting from inserting the piece of polyimide tubing at a depth of approximately 8 mm and parallel across the array transducer elevation aperture. 113 Figure 2.49. Imaging of custom-made tissue-mimicking phantom with inserted piece of polyimide tubing. (A) Piece of polyimide tubing inserted at a depth of 5 mm and parallel across the array transducer elevation aperture. (B) Piece of polyimide tubing inserted at a depth of 8 mm and parallel across the array transducer elevation aperture. 114 CHAPTER 3 FORWARD-LOOKING INTRAVASCULAR IMAGING CATHETER This chapter describes the design, fabrication, and testing of a miniature forward-looking (FL) linear phased array catheter for Intravascular Ultrasound (IVUS) imaging and intraluminal vessel navigation during peripheral vasculature angioplasty procedures. The clinical problem and current standard of care are presented first in order to arrive to the clinical need this catheter is intended to satisfy. The array features 32 piezoelectric elements separated by non-conductive epoxy kerfs, yielding a total active aperture of 0.8 mm in the azimuth direction and 1.0 mm in the elevation direction, with an elevation natural focal depth of 5.0 mm. The array includes non- conductive epoxy backing and two front matching layers. A custom flexible circuit connects the array piezoelectric elements to a bundle of 32 individual micro-coaxial cables (48 AWG) enclosed within a 0.6-m long 8F catheter and terminated with a Verasonics Backshell Connector. 3.1 Clinical Problem: Peripheral Arterial Disease (PAD) Cardiovascular diseases are the number one cause of death globally; more people die annually from cardiovascular diseases than from any other cause (World Health Organization, 2015). Arterial disease (atherosclerosis) is the build-up of cholesterol-laden plaque along the inner walls of arteries, narrowing them (stenosis) and reducing blood flow to different organs and regions of the human body. Peripheral arterial disease (PAD) affects the vessels that carry blood from the heart to the limbs, being more common in the legs than in the arms (Figure 3.1). 115 In the United States, PAD affects 17.6 million people, claims 75,000 lives annually, and results in 100,000 leg amputations (Norgren et al., 2007). Figure 3.1. Peripheral arterial disease (PAD) affects the vessels that carry blood from the heart to the limbs, being more common in the legs than in the arms. (Courtesy of the American Heart Association) 3.2 Current Treatment: Angioplasty Stenotic peripheral arteries are treated by vascular surgeons with a minimally invasive procedure called angioplasty. With an angioplasty procedure, the lumen of a stenotic peripheral artery is restored with the deployment of a balloon and, occasionally, a metal stent. For this purpose, vascular surgeons begin by inserting a needle into the femoral artery in the thigh as a port of access (Figure 3.2). Figure 3.2. Percutaneous access created into the femoral artery with a needle. The guide catheter-guide wire are then navigated through the vasculature to the origin of the blockage, to cross it afterwards, and set the guide wire in position for treatment delivery. (Adapted from the American Medical Center – American Heart Institute) 116 Through this port of access, contrast dye is injected into the artery to show the blood flow and the location of the blockage to be treated using an X-ray medical imaging technique termed angiography. Right next, the vascular surgeon navigates a guide catheter and guide wire to the location of the blockage. Upon reaching the blockage, the vascular surgeon gets across it (Figure 3.3.A). Once the guide wire is across the blockage, the vascular surgeon removes the guide catheter and pushes another catheter with a very small deflated balloon over the guide wire and into the blocked area. The balloon is inflated with air to open the blocked artery (Figure 3.3.B). The vascular surgeon then deflates the balloon completely and removes it. If the artery remains partially blocked, the vascular surgeon pushes another balloon catheter with a collapsed metal stent loaded over the deflated balloon. Once set in place, the balloon is inflated and the stent is opened (Figure 3.3.C). The balloon is then deflated allowing the catheter to be removed. Finally, once blood flow is restored the vascular surgeon removes the guide wire. (A) (B) (C) Figure 3.3. (A) Getting a guide catheter – guide wire across the blockage. (B) Inflating a balloon catheter to open the blocked artery, and (C) Inflating a balloon-stent catheter to deploy the metal stent in cases where the balloon angioplasty still leaves the artery partially blocked. (Adapted from the American Medical Center – American Heart Institute) During the angiography X-ray visualization, the vascular surgeon estimates the size of the artery and the length of the blockage to be treated in order to select the appropriate catheters (and metal stent if needed) that will be used during the angioplasty procedure. While visualizing 117 with X-ray imaging, the vascular surgeon navigates the guide catheter – guide wire through the vasculature and across the blockage, controlling the movement and direction of the guide catheter – guide wire by gently manipulating the end that sits outside the patient through twisting of the guide wire. However, although angiography provides detailed X-ray images of the blood flow in the arteries (e.g. femoral artery) and the location of a blockage, it is imperative to emphasize its incapability to provide images from inside the vessels (Figure 3.4), therefore omitting detailed intraluminal information that is critical for navigating a guide catheter – guide wire through a blockage. On this basis, vascular surgeons necessitate to heavily rely on their haptic skills and experience to avoid poking the vessel wall with the guide wire, an event that can ultimately lead to the dislodgement of pieces of vulnerable plaque (i.e. embolism) or, in the worst of cases, even cutting the walls of the vessels (i.e. partial and total arterial dissections). Figure 3.4. Angiogram showing blood flow in the superficial femoral artery (SFA). The blockage location is clearly visible; however, no images from inside the vessel can be obtained using this technology. Blindly navigating a catheter through the blockage with no intraluminal information (i.e. images) becomes a challenging task for vascular surgeons. (Adapted from WebMD) In addition to angiography, peripheral angioplasty procedures are often performed with another medical imaging technique, Intravascular Ultrasound (IVUS). With IVUS, an ultrasonic transducer mounted on a catheter is capable of producing images from inside of the treated artery 118 (e.g. femoral artery); therefore, it is an excellent means to verify the correct deployment of balloons and stents. However, although IVUS is capable of producing intraluminal images (unlike angiography) it is imperative to emphasize that current state-of-the-art IVUS technology is still not adequate for intraluminal vessel navigation. Due to the side-looking (SL) characteristics of current IVUS catheters (Figure 3.5), obtained intraluminal images show a blind spot at the front that makes it impossible for vascular surgeons to see what is ahead of the catheter, ergo resulting in the aforementioned inadequacy for intraluminal vessel navigation. In few words, with current imaging technologies intraluminal vessel navigation is being done blindly. Figure 3.5. State-of-the-art SL-IVUS catheters: (A) Rotating single element catheter, (B) Circular array catheter. (SL-IVUS blind spot image adapted from Volcano Corporation) As shown in Figure 3.6, out of the 300,000 peripheral interventions performed in the United States, in up to 40% of the cases the vessels present chronic total occlusions (CTOs) (Boguszewski et al., 2010) with blockages that can be as long as 10 cm or more, sometimes even the whole length of the vessel as it has happened in cases involving the superficial femoral artery (Ryan, 2015). Due to the length of peripheral blockages, the probability of causing an embolism (dislodging small pieces of plaque) can be as high as 85%, as a result of the blind navigation of a 119 catheter with its guide wire through a blockage. Because of an embolism, the artery could become blocked not only in one, but also in two or more locations as a result of the embolus traveling further along the vessel. It is because of this reason that vascular surgeons nowadays employ a technique called “sub-intimal angioplasty”. With this technique, the probability of causing an embolism can get reduced to approximately only 3% (Woo, 2015). However, it is imperative to mention that such technique involves cutting on purpose the tunica intima of the vessel in order to create a false lumen in the tunica media through which the catheter and its guide wire can circumvent the critical regions of the blockage to avoid causing an embolism. Nevertheless, the use of this technique represents an incidence of up to 80% of the cases in which the vessel suffers partial dissection (Han, 2015). Additionally, the long-term repercussions of causing these lesions on purpose have been linked to strong inflammatory responses, especially when the technique is not well executed and the cuts reach the tunica adventitia. This has resulted in up to 54% of the cases presenting total vessel dissection (Krishnan et al., 2012). Figure 3.6. Incidence of adverse events in peripheral vascular interventions performed in the United States. 120 As shown in Figure 3.7, the presence of adventitial cuts leads to a restenosis frequency of up to 97%, whereas their absence decreases it to 15%. Since restenosis means that the treated vessel becomes occluded again, the patient (and the vascular surgeon) must deal with the drawbacks of a compulsory second angioplasty procedure (Krishnan et al., 2012). Figure 3.7. Effect of adventitial cuts on restenosis frequency for peripheral interventions via sub-intimal angioplasty. In a study involving 102 patients, 54% of them presented adventitial cuts with a concomitant restenosis frequency of 97%. (Krishnan et al., 2012) In conclusion, on the one hand, medial cuts incidence is high (up to 80%) as a result of applying the sub-intimal angioplasty technique in order to achieve a low embolism incidence (approx. 3%). This technique entails a significant adventitial cuts incidence (up to 54%), with a concomitant high restenosis frequency (up to 97%). On the other hand, if sub-intimal angioplasty is not to be used, in order to avoid medial and adventitial cuts, embolism will most likely occur (up to 85% incidence) as a result of the catheter with its guide wire being blindly pushed through the blockage. 121 3.3 Clinical Need To summarize the clinical need, vascular surgeons need a means to see the path and structures ahead of a guide wire in order to perform faster, safer, and easier intraluminal vessel navigation during peripheral angioplasty interventions of highly stenosed lesions. The proposed forward-looking intravascular ultrasound (FL-IVUS) phased array catheter will satisfy this need by providing real-time, high-resolution images still unobtainable with current angiography and IVUS technologies of the path and structures in front of the IVUS catheter (Figure 3.8). Electronic beam steering, inherent to phased arrays, together with minimal mechanical maneuvers will enable visualization of the whole 3-D space allowing vascular surgeons to traverse occlusions more easily and safely in letting them move the catheter in the right direction before the guide wire can make any contact with structures in the front. This device will lead to more successful and shorter peripheral angioplasty interventions. Figure 3.8. FL-IVUS phased array catheter is capable of imaging stenosed areas in front of the catheter with no blind spots. 3.4 Array Transducer Design The array design incorporates the required features for peripheral vessel navigation: Array type, aperture size, frequency, orientation, and catheter packaging size. A forward- 122 looking, linear phased array is needed in order to image all the structures in front of the catheter. In practice, vascular surgeons could use this FL-IVUS catheter right after creating the percutaneous access in order to navigate the guide wire through the peripheral vasculature. However, to expedite navigation of blockage-free regions, vascular surgeons could still rely on angiography until the catheter reached the blocked area. Upon reaching, vascular surgeons would rely on the images provided by the FL-IVUS catheter to traverse the blockage faster, more easily, and safely. The FL-IVUS phased array catheter was modeled via the computer-aided design (CAD) software SolidWorks (Dassault Systèmes SolidWorks Corp., Waltham, MA). As shown in Figure 3.9, the array acoustic stack is enclosed within a piece of 8F polyimide tubing having an internal diameter of 2.5 mm together with a flexible circuit that excites each one of the piezoelectric elements of the 2-2 composite and clamps this one on its left and right ends to provide ground connection. The design of the flexible circuit allows connecting and fitting a bundle of individual micro-coaxial cables (48 AWG) inside of an 8F polytetrafluoroethylene (PTFE) catheter. Figure 3.9. CAD model of FL-IVUS linear phased array enclosed in protective 8F catheter. 123 3.4.1 Array Acoustic Stack The array center frequency was chosen to be 30 MHz to achieve axial and lateral resolutions necessary to reliably visualize the whole space in front of the catheter and assist in the navigation of the guide wire through an artery blockage. The final array geometry and materials were determined considering the concepts and design rules presented in sections 1.3 Ultrasound Transducers and 1.4 Ultrasound Arrays, as well as simulation results and fabrication process limitations. Lead magnesium niobate-lead titanate (PMN-PT) (CTS Corporation, Bolingbrook, IL) single crystal was chosen as the piezoelectric material due to its high dielectric constant and low dielectric loss, which make it ideal for high-sensitivity transducers with small- aperture size (Chen et al., 2013). The array 2-2 composite consists of thirty-two 19- m-wide piezoelectric elements separated by 6- m-wide kerfs filled with non-conductive epoxy (Epo-Tek 301) (Epoxy Technology, Inc., Billerica, MA) to yield an array with a 25- m-wide pitch (0.5water pitch for the suppression of undesirable grating lobes during beam steering), a total active azimuth aperture of 0.8 mm, and a total active elevation aperture of 1 mm; features that represent an important and considerable progress in the area of miniature FL catheter array transducers where most current designs employ more than two times larger pitches (~65 m) with concomitant larger element-kerf sizes and larger imaging apertures (Stephens et al., 2009). The array acoustic stack includes a backing layer of non-conductive epoxy (Epo-Tek 301) mixed with 43.2 wt% liquid plasticizer (LP-3) (Structure Probe, Inc., West Chester, PA) and two front matching layers: 1. Conductive silver-loaded epoxy consisting of 2,3- m silver powder (Aldrich Chemicals Co., St. Louis, MO) with mixed epoxy (Insulcast 501 + Insulcure 9) (ITW Polymers Coatings North America, Montgomeryville, PA) and 2. Vapor deposited polymer (Parylene C) 124 (Specialty Coating Systems, Inc., Indianapolis, IN). The array acoustic stack final design parameters and material properties are shown next in Table XII and Table XIII, respectively. Table XII. FL-IVUS linear phased array design parameters. Design center frequency 30 MHz Number of elements 32 Composite configuration 2-2, linear Pitch (0.5 water) 25 m Element width 19 m Kerf width 6 m Elevation aperture 1 mm Azimuth aperture 0.8 mm Elevation focal distance (natural) 5 mm Piezoelectric material PMN-PT 27 m thick Kerf filler material Non-conductive epoxy (Epo-Tek 301) 27 m thick Backing material Non-conductive epoxy (Epo-Tek 301) with 43.2 wt% liquid plasticizer (LP-3) 2 mm thick 1 st Matching layer material Conductive epoxy (2,3- m silver epoxy) 25 m thick 2 nd Matching layer material Vapor deposited polymer (Parylene C) 15 m thick Table XIII. Material properties of the FL-IVUS linear phased array acoustic stack. Material PMN-30%PT single crystal 1 Epo-Tek 301 2 Epo-Tek 301 + LP-3 (43.2 wt%) 3 2,3- m silver epoxy 4 Parylene C 5 Density, ρ (kg/m 3 ) 7,800 1,150 1,230 3,860 1,100 Acoustic longitudinal velocity, c (m/sec) 4,600 2,650 2,450 1,900 2,350 Acoustic impedance, Z (MRayl) 35.88 3.05 3.01 7.33 2.59 Piezoelectric strain constant, d33 (C/N) 1,500 x 10 -12 - - - - Free dielectric constant, K к ~5,000 - - - - Clamped dielectric constant, K ε 800 - - - - 125 Dielectric loss tangent, tanδe 0.005 - - - - Electromechanical coupling coefficient, kt 0.58 - - - - 1. CTS Corporation, Bolingbrook, IL 2. Epoxy Technology, Inc., Billerica, MA 3,4. Resource Center for Medical Ultrasonic Transducer Technology at the University of Southern California 5. Specialty Coating Systems, Inc., Indianapolis, IN 3.4.2 Electrical Interconnect – Flexible Printed Circuit One critical challenge of fabricating miniaturized high-frequency catheter arrays is creating an electrical interconnect solution. For the array under discussion, the element width and pitch, together with the number of elements and the available space inside of an 8F catheter (2.5 mm internal diameter), required designing a high-density flexible circuit that pushes the limits of current technology for flexible printed circuitry production in terms of electrical trace width and pitch. The design endeavor was a lengthy process that required dealing with six different flexible circuitry manufacturers and drafting over 30 design iterations in a timespan of over 10 months until converging to the final solution. Figure 3.10 shows the CAD schematics (AutoCAD, Autodesk, Inc., San Rafael, CA) of the flexible circuit design whose central area lies right below the 2-2 composite to provide its 32 piezoelectric elements with electrical connection via thirty two 12- m-wide signal traces with a 25- m-wide pitch, divided into two groups of 16 signal traces each. Each group further fans out into 25- m-wide signal traces with a 101- m-wide pitch, each of which will then be connected to an individual 48 AWG micro-coaxial cable. 126 Figure 3.10. CAD schematics of flexible circuit design. 3.4.3 Electrical Interconnect – Cabling The micro-coaxial bundle features 32 individual 48 AWG micro-coaxial cables (Hitachi Metals, Ltd., Tokyo, Japan), whose properties are shown below in Figure 3.11. 3. PFA = Perfluoroalkoxy Figure 3.11. High-capacitance 48 AWG micro-coaxial cable. (Courtesy of Hitachi Metals, Ltd.) 127 In order to determine the optimal micro-coaxial cable length to be used, scattering parameters (S-Parameters) were evaluated to achieve the best performance in terms of reflection and transmission. S-Parameters are employed to represent high-frequency networks in accordance to direct measurements of reflected and transmitted waves. They provide a complete description of the network regarding how the incident waves are related to the reflected and transmitted waves (Figure 3.12). S21 S11 S12 S22 2 Port Network + - V1 V1+ V1- + - V2 V2+ V2- Figure 3.12. 2-Port Network representation of a micro-coaxial cable for estimating its scattering parameters: Reflection and Transmission Coefficients. S-Parameters are defined as (Pozar, 2012): 2 1 22 12 21 11 2 1 V V S S S S V V (3.1) 1 1 11 V V S where 0 2 V It defines the reflection at port 1 due to impedance mismatch. 1 2 21 V V S where 0 2 V It defines the gain (loss) of system from port 1 to port 2. 2 1 12 V V S where 0 1 V It defines the gain (loss) of system from port 2 to port 1. 2 2 22 V V S where 0 1 V It defines the reflection at port 2 due to impedance mismatch. 128 The S-Parameters of the 48 AWG micro-coaxial cable were measured at three different lengths (0.6, 1, and 1.5 m) with an Agilent E5072A Network Analyzer (Agilent Technologies, Santa Clara, CA). Due to reciprocity of the system, S11 = S22 and S21 = S12. Figures 3.13 and 3.14 show the effect of micro-coaxial cable length on the S-Parameters. Figure 3.13. Reflection coefficient of the 48 AWG micro-coaxial cable at 0.6, 1.0, and 1.5 m. Figure 3.13 shows that at 30 MHz the reflection coefficient of the 48 AWG micro-coaxial cable generates less reflection at the input and output at 0.6 and 1.5 m cable lengths; therefore having less oscillatory effect on the transmitted and received signal. 129 Figure 3.14. Transmission coefficient of the 48 AWG micro-coaxial cable at 0.6, 1.0, and 1.5 m. Figure 3.14 shows that at 30 MHz the transmission coefficient of the 48 AWG micro- coaxial cable decreases with increasing cable length. In other words, the loss of transmitted signal increases with increasing cable length. Therefore 0.6 m cable length delivers higher power to the drive the array transducer. In conclusion, both S11 and S21 indicate that 0.6 m cable length is the most efficient option of the three lengths tested and exerts the lowest attenuation on the signal for this first generation of the FL-IVUS phased array catheter. Therefore, we chose 0.6 m length for each individual 48 AWG micro-coaxial cable. 3.4.4 Electrical Interconnect – Verasonics Backshell Connector Kit The array is to be driven by a Verasonics Vantage 128 System (Verasonics, Inc., Kirkland, WA). In order to take advantage of the system capabilities and to avoid any possible incompatibility drawback, a Backshell Connector Kit (Verasonics, Inc., Kirkland, WA) was 130 procured to connect the array to the system. The connector kit (Figure 3.15) includes: 4 micro- coax termination boards (plus 2 spares) of which only 1 is to be used by the array being presented; backshell left and right halves, 1 ZIF (zero-insertion force) connector PCB (printed circuit board), 1 Cannon DL-260 connector plug and locking handle, 2 tuning PCBs, and assembly documentation. Overall, to assemble the kit, the micro-coaxial cables must first be soldered to 1 termination board, which handles all the required 32 signal channels. Additionally, should tuning components be necessary, these also have to be soldered to the corresponding tuning board. The termination board is then plugged into the tuning board, and this, into the ZIF connector PCB. The resulting subassembly is then placed and fitted inside one backshell half and finally the other backshell half is secured in place with the included screws. A more detailed description of the Backshell Connector Kit assembly for the array under discussion is included ahead in section 3.6 Array Transducer Fabrication. Figure 3.15. Verasonics Backshell Connector Kit. (Courtesy of Verasonics, Inc.) 131 3.4.5 Verasonics Imaging System The Verasonics Vantage 128 System, shown in Figure 3.16, counts with 128 transmit (Tx) and 128 receive (Rx) channels that are independently controlled, of which only 32 Tx / 32 Rx channels will be used by the array under discussion. On Tx mode, the system is to be operated at a programmable center frequency of 30 MHz with pulse amplitudes of 100 Vpp or lower. Even though the array has been designed with a 0.5λwater pitch, the system also counts with per-channel Tx apodization using pulse-width modulation to assist in the suppression of undesirable grating lobes during beam steering. On Rx mode, the system counts with real-time access to RF data for each individual channel should it be necessary (e.g. verify that an array element is indeed receiving returning echoes and generating corresponding electrical signals) and a programmable anti-aliasing filter. The technical specifications of the Verasonics Vantage 128 System are summarized below in Table XIV. Figure 3.16. Verasonics Vantage 128 System. 132 Table XIV. Verasonics Vantage 128 System technical specifications. Transmit (Tx) All transmit and receive channels are independently controlled Standard frequency range: 0.5 MHz to 20 MHz o (2 – 42 MHz with High Frequency configuration) o (100 KHz – 1700 KHz with Low Frequency configuration) Time delay resolution: 4.0 nsec. Programmable pulser amplitude, 3 to 190V pp Tri-state drive: + high voltage, – high voltage and ground Per-channel, programmable center frequency, pulse width (pulse duty cycle), pulse length, polarity and delay Per-channel transmit apodization using pulse width modulation Max burst length (without Extended Transmit option) of 20 cycles at transmit center frequency Power limit, single channel: up to 100 Watts peak, 8 Watts average (into 50 Ohms) Receive (Rx) Real-time access to RF data for each channel Standard frequency range: 0.5 – 27 MHz o (1 – 50 MHz with High Frequency configuration) o (100 KHz – 1700 KHz with Low Frequency configuration) Programmable anti-aliasing filter cutoff: 10, 15, 20 and 30 MHz, and in addition, 35 and 50 MHz with High Frequency configuration 14 bit A/D converters with programmable sample rate up to 62.5 MHz Two independent, user-programmable, symmetrical RF data digital (23 tap and 41 tap) 3.5 Array Transducer Simulation Modeling The final design of the FL-IVUS phased array catheter was defined via the finite-element analysis (FEA) software PZFlex (Weidlinger Associates, Inc., Cupertino, CA). FEA modeling is capable of providing an accurate prediction of the performance of a high-frequency array, therefore reducing the number of time-consuming prototype fabrication runs (Chen et al., 2014). FEA modeling was performed following the methodology previously presented in section 2.5 Array Transducer Simulation Modeling for the SL-ERUS linear phased array. Likewise, the 133 array performance was evaluated and optimized via simulations that included: 1. Pulse-echo response and electrical impedance, 2. Electrical and acoustical crosstalk, and 3. Pressure wave generation and propagation, pressure field, and axial/lateral beam profiles. Input array parameters are listed in Table XII and the 2D model is shown below in Figure 3.17. Figure 3.17. 2D model of the FL-IVUS phased array catheter with its x-y coordinates (symmetry boundary condition applied). 3.5.1 Pulse-Echo Response and Electrical Impedance Figure 3.18 shows the results obtained for one of the 32 piezoelectric elements of the FL- IVUS phased array catheter (element 17) excited by a single-cycle sinusoid (pulsed wave excitation) with an amplitude of 60 Vpp at the design center frequency. The left subfigure shows 134 the voltage history generated by the received echo (μs vs. mV) and its frequency spectrum (MHz vs. dB), rendering a peak-to-peak sensitivity of 33.6 mV, an axial resolution of 54.75 μm (equation 1.20), a -6 dB center frequency of 27.1 MHz, and a -6 dB fractional bandwidth of 41.2%. The right subfigure shows the piezoelectric element electrical impedance magnitude (MHz vs. Ω) and phase (MHz vs. deg), rendering an electrical impedance magnitude of 109.5 Ω with a phase of -42.96° at 30 MHz and an electromechanical coupling coefficient (equation 1.13) of 0.67, close to the reported value of 0.58 for bulk PMN-PT single crystal by CTS Corporation (CTS Corporation, Bolingbrook, IL). Figure 3.18. Results obtained for representative piezoelectric element 17. Left: Voltage generated by received echo and its frequency spectrum. Right: Electrical impedance magnitude and electrical impedance phase. 3.5.2 Electrical and Acoustical Crosstalk In order to get the crosstalk across a range of frequencies, a 5-Volt, five-cycle sinusoid was applied within a frequency range from 10 to 50 MHz in steps of 2 MHz. These simulations were executed on element 16 with respective adjacent elements 15, 14, 13, and 12 as well as on element 8 with respective adjacent elements 7, 6, 5, and 4. Afterwards, the crosstalk for first adjacent elements, second adjacent elements, third adjacent elements, and fourth adjacent 135 elements were averaged and plotted as relative amplitudes. As shown next in Figure 3.19, maximum crosstalk levels at 30 MHz were: -18.5 dB for the first adjacent element, -24.5 dB for the second adjacent element, -24.3 dB for the third adjacent element, and -26.1 dB for the fourth adjacent element. Although these values are higher than the recommended < -35 dB for linear phased arrays (Table IV), one has to consider the small aperture of the linear phased array with its concomitant narrow kerfs. Should it be necessary, in order to bring the crosstalk down to < -35 dB we could include attenuating materials to the kerf filler (e.g. adding liquid plasticizer, LP-3, to the non-conductive epoxy, Epo-Tek 301), like it is done in the case of the backing layer. Figure 3.19. Computed crosstalk from 10 to 50 MHz. Maximum crosstalk levels at 30 MHz were: -18.5 dB for the first adjacent element, -24.5 dB for the second adjacent element, -24.3 dB for the third adjacent element, and -26.1 dB for the fourth adjacent element. 3.5.3 Pressure Field and Axial/Lateral Beam Profiles Figure 3.20 shows a PZFlex screenshot of the resulting pressure field in the loading medium (Pa); notice how the pressure reaches its maximum at the elevation natural focal point of 136 5 mm. Figure 3.21.A shows the axial beam profile in the loading medium along the center line of the array 2D model (mm vs. dB); notice how the axial intensity (relative pressure) oscillates in the near field, reaches its maximum at the elevation natural focal point of 5 mm (transition point), and decays in the far field. Figure 3.21.B shows the lateral beam profile in the loading medium across the elevation natural focal plane (mm vs. dB), with a side-lobe level (SLL) of -26.5 dB, and with the -6 dB beam width yielding a lateral resolution of 250 μm. Figure 3.20. Pressure field in the loading medium of the FL-IVUS phased array catheter under continuous wave excitation (100-cycle sinusoid with an amplitude of 100 Vpp at the design center frequency). Total loading medium length is 10 mm. Figure 3.21. (A) Axial beam profile in the loading medium along the centerline of the array. (B) Lateral beam profile in the loading medium across the elevation natural focal plane at 5 mm in front of the array. Table XV below comprehensively summarizes the FL-IVUS phased array catheter design parameters and simulated performance results. 137 Table XV. FL-IVUS phased array catheter design parameters and simulated performance. Design Parameters Design center frequency 30 MHz Number of elements 32 Composite configuration 2-2, linear Pitch (0.5 water) 25 m Element width 19 m Kerf width 6 m Elevation aperture 1 mm Azimuth aperture 0.8 mm Elevation focal distance (natural) 5 mm Piezoelectric material PMN-PT 27 m thick Kerf filler material Non-conductive epoxy (Epo-Tek 301) 27 m thick Backing material Non-conductive epoxy (Epo-Tek 301) with 43.2wt% liquid plasticizer (LP-3) 2 mm thick 1 st Matching layer material Conductive epoxy (2,3- m silver epoxy) 25 m thick 2 nd Matching layer material Vapor deposited polymer (Parylene C) 15 m thick Simulated Performance -6 dB center frequency 27.1 MHz -6 dB fractional bandwidth 41.2 % -6 dB / -20 dB pulse length 1 78 / 146 nsec -6 dB lateral resolution 2 250 μm Side-lobe level (SLL) 2 -26.5 dB Peak-to-peak sensitivity 1 33.6 mV Axial resolution 2 54.75 μm Electrical impedance 1 109.5 , -42.96° @ 30 MHz Electromechanical coupling coefficient 1 0.67 Electrical and acoustical crosstalk < -18.5 dB @ 30 MHz 1. For representative array element 16 2. At elevation natural focus 3.6 Array Transducer Fabrication Fabricating ultrasound array transducers is a challenging task especially when aperture miniaturization and catheter housing are involved. Furthermore, issues such as epoxy bond 138 thickness and uniformity, crack-free piezoelectric crystal and electrode layers, as well as material losses play a primordial role in determining the performance of a device at high frequencies. This section describes the techniques and materials used to fabricate the FL-IVUS phased array catheter. The array 2-2 composite was fabricated via Deep Reactive Ion Etching (DRIE). The custom flexible circuit was used to connect the array elements to the high-capacitance coaxial cable assembly consisting in a 0.6-m long bundle of 32 individual micro-coaxial cables (48 AWG), which in turn were terminated onto one micro-coax board inside of the proprietary Verasonics Backshell Connector. The array acoustic stack was enclosed within a 2.5-mm long piece of 8F polyimide tubing and the cable assembly was fit inside of a 0.6-m long piece of 8F catheter fabricated out of biocompatible polytetrafluoroethylene (PTFE). 3.6.1 Array Acoustic Stack (2-2 Composite + 1 st Matching Layer) The array PMN-PT 2-2 composite was fabricated by CTS Corporation (CTS Corporation, Bolingbrook, IL) using a Deep Reactive Ion Etching (DRIE) technique optimized for piezoelectric materials. This etching technique was previously used successfully for a couple of 60 MHz ultrasound arrays and we modified it for the array under discussion (Liu et al., 2013) (Cummins et al., 2016). The array composite fabrication process begins with bulk PMN-PT single crystal material and ends with individual array acoustic stacks (2-2 composite + 1 st matching layer) ready to be bonded to their respective flexible circuits. DRIE was chosen in order to achieve the small kerf dimensions required since it is an anisotropic etching process that is useful for creating microstructures out of bulk materials with especially high aspect ratios and steep sidewall angles. DRIE is a micromachining process originally developed for dynamic random access memory (DRAM) fabrication applications. 139 DRIE uses heated plasma to etch into bulk material using a photolithographic developed pattern of a hard metal such as nickel to serve as the etching mask. The high resolution of photolithography combined with the capability of DRIE to etch through materials with thicknesses larger than 20 µm makes it an ideal choice to create small kerfs in high-frequency composite arrays (Cummins et al, 2016). The first step consisted in lapping and polishing by hand a bulk piece of PMN-PT single crystal grown by CTS Corporation to ensure coplanarity of the top and bottom surfaces until a thickness of 1 mm was reached. Next, a nickel mask was patterned onto the bulk material in such a way that nickel covered all of the material except for the regions intended to be kerfs. Since the final array size is only 0.98 mm x 1.46 mm and the piezoelectric bulk material is 1.5 cm x 1.5 cm, a 7 x 7 matrix of arrays can be patterned during a single batch, enabling 49 arrays to be fabricated from each bulk material sample of PMN-PT single crystal. This helped to account for the potentially low yield of undamaged and useable arrays at the end of the fabrication process. Additionally, we were interested in developing this batch fabrication method because it has much more potential as a reliable commercial fabrication approach than traditional array fabrication processes, which allow for only one array to be fabricated by hand at a time. After the 49 arrays were patterned with the nickel mask, the sample was run through the DRIE process to etch down 35 µm into the bulk material. Etching past the designed composite thickness of 27 µm was necessary to allow for several microns of material to be removed in the subsequent lapping fabrication steps. We chose to investigate 3 different composite designs for the composite posts. Each design had the same element and kerf width in the azimuth direction but each element was sub-etched in the elevation direction at pitches of 50.3 µm, 100.5 µm and 140 201 µm, each with a sub-element kerf width of 5 µm. This sub-element approach was advised and etched by CTS Corporation to prevent a crack from propagating along an entire element and also to mitigate the risk of future element electrodes breaking along the narrow composite pillars due to strain. On this basis, the etched sub-elements of each composite pillar could have flexible non-conductive epoxy, Epo-Tek 301, inserted at regular intervals that could allow for expansion or contraction of the piezoelectric material without all of the strain being transferred to the thin chrome / gold electrode later on patterned on the element. Figure 3.22 shows the etching mask used to create the composites. To improve the array fabrication yield, small 30 µm x 30 µm posts in a 1-3 composite arrangement were patterned between each array to break up the solid pieces of piezoelectric material and prevent cracks from propagating from one array to another. After DRIE processing and filling in the etched kerfs with Epo-Tek 301 epoxy, the bulk sample was then lapped down to expose the top of the piezoelectric elements. Once the array elements were lapped to a smooth matte finish, the bulk sample was shipped by CTS Corporation to our group, the Resource Center for Medical Ultrasonic Transducer Technology at the University of Southern California. Figure 3.23 shows three individual arrays, one for each of the three aforementioned sub-element pitch lengths. Upon receipt, the bulk sample was additionally lapped with 9-μm diamond suspension (Buehler-Illinois Tool Works, Inc., Lake Bluff, IL) to remove any excess of kerf filler epoxy and make sure that the whole first lapped side of the sample was completely flat. 141 Figure 3.22. (A) DRIE nickel mask for all 49 arrays with 2-2 composite pattern nested within a 1-3 composite pattern to prevent crystal fracture propagation in the sample. (B) Single array with 32-element aperture size of 0.8 mm x 1 mm. (C) 3 distinct individual sub-element lengths for each array ceramic pillar to prevent electrode strain that could result in open channel circuits. 142 Figure 3.23. Micrographs of three individual arrays as received from CTS Corporation. Afterwards, the sample was mechanically cleaned by applying acetone, reagent alcohol, and Alconox detergent (Alconox, Inc., White Plains, NY) with a cotton swab and any residues left were removed with deionized water. The lapped and mechanically cleaned side of the sample was then plasma cleaned (argon 25 sccm, 30 watts, 185 sec) to activate the exposed surface of the piezoelectric elements for electroplating sputtering of their active electrodes having a thickness of 500 Å of chrome and 2,000 Å of gold. Since electroplating sputtering actually covered the whole top surface of the sample, the active electrodes were patterned on the piezoelectric elements by removing gold and chrome from the epoxy kerfs with a cotton swab dampened in reagent alcohol, taking advantage of the weak adhesion of chrome and gold to epoxies. The sample was then flipped over and lapped down to its final design thickness. 2,000- grit sandpaper together with 5-μm aluminum oxide (Al₂O₃) powder (Buehler-Illinois Tool Works, Inc., Lake Bluff, IL) were used to ensure a smooth, matte finish and 9-μm diamond suspension lapping was done once again to remove any excess of kerf filler epoxy. The exposed second side of the sample underwent the same process of mechanical and plasma cleaning, together with electroplating sputtering now of the common ground electrode, reason why there was no electrode patterning on this occasion. 143 The 1 st matching layer was then prepared by adding 4.5 gr of 2,3- m silver powder (Aldrich Chemicals Co., St. Louis, MO) to 1.25 gr of mixed non-conductive epoxies: Insulcast 501 and Insulcure 9 (ITW Polymers Coatings North America, Montgomeriville, PA). The resulting mixture was then degassed to avoid the presence of air bubbles, cast on top of the common ground electrode, and centrifuged at 3,000 RPM for 15 mins to distribute it evenly. After curing overnight in a dry-environment nitrogen box and post-curing in a conventional oven for 2 hrs at 45 °C, the 1 st matching layer was lapped down to its final design thickness and the resulting acoustic stack (2-2 composite + 1 st matching layer) was mechanically diced (0.5 mm/sec feed rate and 30,000 RPM spindle speed) to final dimensions (Figure 3.24). Figure 3.24. Acoustic stack comprising the PMN-PT 2-2 composite and 1 st matching layer. Active and ground electrodes were sputtered on the bottom and top sides of the composite, respectively. Both layers were lapped down to final design thicknesses and the resulting stack was mechanically diced to final dimensions. 144 3.6.2 Flexible Printed Circuit The flexible circuit was fabricated by Cicor (Cicor Advanced Microelectronics, Ulm, Germany). Figure 3.25 shows a physical sample of the flexible circuit consisting of a 12.5-μm- thick base dielectric polyimide layer (Kapton) (DuPont, Wilmington, DE), which features signal traces on the front and a common ground electrode on the back. All signal traces and ground electrodes were patterned over a 2-μm-thick copper layer and finished with a flash of gold. The flexible circuit central area provides the 2-2 composite 32 piezoelectric elements with electrical connection via thirty two 12- m-wide signal traces with a 25- m-wide pitch, divided into two groups of 16 signal traces each. Each group further fans out into 25- m-wide signal traces with a 101- m-wide pitch. Figure 3.25. Flexible printed circuit for the FL-IVUS phased array catheter. 3.6.3 Bonding the Array Acoustic Stack and Backing Layer to the Flexible Printed Circuit After mechanical dicing to final dimensions, both sides of the acoustic stack were cleaned by applying trichloroethylene, acetone, and reagent alcohol with a cotton swab. Likewise, the front side of the flexible circuit was cleaned with a cotton swab dampened in reagent alcohol. Afterwards, a thin line of Epo-Tek 301 epoxy was applied with a foam swab at the center of the 145 flexible circuit (2-2 composite area). Another foam swab was used then to smear a thin layer of Epo-Tek 301 epoxy all over the surface of the 2-2 composite (bottom side of the acoustic stack). As shown in Figure 3.26, the acoustic stack was afterwards placed on top of the flexible circuit (left subfigure) visually making sure that the piezoelectric elements were correctly aligned with the signal traces through a previously etched window on the backside ground electrode of the flexible circuit (center subfigure), and pressure was applied using a “C” clamp (right subfigure). The resulting sub-assembly was cured overnight in a dry-environment nitrogen box and then post-cured in a conventional oven for 2 hrs at 45 °C. Figure 3.26. Bonding the array acoustic stack to the flexible circuit. Left: Array acoustic stack placed on top of the flexible circuit. Center: Piezoelectric elements correctly aligned with the signal traces of the flexible circuit. Right: Pressure applied with a “C” clamp to bond the resulting sub-assembly. The next step consisted in bonding the backing layer to the backside of the flexible circuit. For this purpose, the backside of the flexible circuit was cleaned with a cotton swab dampened in reagent alcohol. Afterwards, a thin line of Epo-Tek 301 epoxy was applied with a foam swab at the center of the flexible circuit (backside of 2-2 composite area) and another foam swab was used then to smear a thin layer of Epo-Tek 301 epoxy all over the bonding surface of the backing layer. The backing layer was afterwards placed on top of the flexible circuit backside 146 and pressure was once again applied with a “C” clamp. Likewise, the resulting sub-assembly was cured overnight in a dry-environment nitrogen box and then post-cured in a conventional oven for 2 hrs at 45°C. Figure 3.27.A shows the resulting sub-assembly with the array acoustic stack and backing layer bonded to the flexible circuit. In order to connect the common ground electrode on the backside of the flexible circuit to the front side of the 1 st matching layer, the two grounding flaps featuring metallized thru-vias were bent and then bonded to the front side of the 1 st matching layer using conductive epoxy, E-Solder 3022 (Von Roll Isola USA, Inc., Schenectady, NY). Afterwards, the two branches of the flexible circuit were bent with a radius of 500 μm and bonded to the sides of the backing layer using Epo-Tek 301 epoxy (Figure 3.27.B). More E-Solder 3022 epoxy was then added along the sides of the two branches of the flexible circuit and down to their fan-out areas in order to increase the contact area with the common ground electrode on the backside (Figure 3.27.C). Finally, the resulting sub-assembly was enclosed within a 2.5-mm long piece of 8F polyimide tubing and the gaps between these two were filled with Epo-Tek 301 epoxy, which also served as protection for the subsequent fabrication steps (Figure 3.27.D). According to flexible circuit design criteria (Finstad, 2008) (IPC, 2016), we could have used a 200-μm bending radius for the branches of the flexible circuit in order to bring the size of the polyimide tubing, and consequentially of the PTFE catheter, down to 6F. However, for this first generation of the FL-IVUS phased array catheter we conservatively chose the aforementioned 500-μm bending radius to absolutely avoid breaking any of the signal traces. Figure 3.28 shows a cross-sectional image revealing the diverse components of the sub-assembly and a detail view of the bending of one branch of the flexible circuit, making sure there were no broken signal traces. 147 Figure 3.27. (A) Array acoustic stack and backing layer bonded to the flexible circuit. (B) Bending the two branches of the flexible circuit and bonding them to the sides of the backing layer with non-conductive Epo-Tek 301 epoxy. (C) Resulting sub-assembly featuring increased contact area with the common ground electrode on the backside of the flexible circuit via conductive E-Solder 3022 epoxy. (D) Resulting sub-assembly surrounded by non-conductive Epo-Tek 301 epoxy and enclosed within a 2.5-mm long piece of 8F polyimide tubing. 148 Figure 3.28. (A) Cross-sectional image revealing the diverse components of the array acoustic stack and backing layer bonded to the flexible circuit. (B) Detail view of the bending of one branch of the flexible circuit, making sure there were no broken signal traces. 3.6.4 Cable Assembly Connection to the Flexible Printed Circuit This section was possibly the most challenging one of the FL-IVUS probe fabrication process mainly due to the fragility of the miniature 48 AWG micro-coaxial cables (e.g. 171-μm- diameter jacket and 39-μm-diameter inner conductor), as well as the required tight soldering pitch of only 101 μm. The first step consisted in assembling the 0.6-m long bundle of 32 pieces of 48 AWG micro-coaxial cable and enclosing it inside of an 8F catheter. For this purpose, the micro-coaxial cable was cut into thirty-two 0.6-m long pieces with their ends subsequently codified with colors and a corresponding number, respectively. Afterwards, all the 32 pieces were carefully aligned on the color end and glued together with Loctite 430 (Henkel AG & Co., Düsseldorf, Germany) inside of a 1-cm long piece of 6F polytetrafluoroethylene (PTFE) tubing to which a 1-m long 1F guide wire was hooked into (Figure 3.29.A). All the 32 pieces of micro- coaxial cable were then manually pulled through a 0.55-m long piece of 8F PTFE tubing, and upon exiting, the 1-cm long piece of 6F PTFE tubing was cut off in order to immediately proceed 149 to number the other end of each micro-coaxial cable according to their color code (Figure 3.29.B). Figure 3.29. Cable assembly and packaging process. (A) All thirty-two 0.6-m long pieces of 48 AWG micro-coaxial cable carefully aligned and glued together inside of a 1-cm long piece of 6F PTFE tubing to which a 1F guide wire was hooked into. (B) All 32 pieces of micro-coaxial cable pulled through a 0.55-m long piece of 8F PTFE tubing, piece of 6F PTFE tubing cut off, and other end of each micro-coaxial cable numbered. The next step consisted in manually stripping each piece of 48 AWG micro-coaxial cable according to the specifications shown in Figure 3.30; a process that proved to be laborious and intensive since this fine-gauge micro-coaxial cable can become easily damaged (e.g. bent inner conductor, broken insulation, or broken inner conductor). Automatic, industrial-grade stripping services were sought (Laser Wire Solutions Ltd., Pontypridd, UK) but the scarcity of service providers for this fine gauge as well as the high lead time and costs involved (e.g. tooling and fixture development, process setup and refinement, and man-hours) resulted in our group having to forgo such an option. 150 Figure 3.30. Stripping specifications of 48 AWG micro-coaxial cable. Afterwards, each single piece of 48 AWG micro-coaxial cable was carefully placed, aligned, and secured onto the flexible printed circuit. Once secured, conductive E-Solder 3022 epoxy was manually applied to cold-solder each 39-μm-diameter inner conductor to its respective 25-μm-wide signal trace; a process that resulted lengthy and challenging as extreme caution had to be taken in order to avoid creating electrical shorts between consecutive signal traces having an extremely tight pitch of only 101 μm. After completing the first side with the first 16 connections, more E-Solder 3022 epoxy was then manually applied to cold-solder all the exposed outer conductors with each other (Figure 3.31.A), and non-conductive Epo-Tek 301 epoxy was manually applied all over the flexible circuit branch to pot and protect the connections created (Figure 3.31.B). The same process was repeated for the second side with the remaining 16 connections. The next step consisted in manually applying more E-Solder 3022 151 epoxy to the backside of both branches of the flexible circuit as well as to the backside of the backing layer in order to create a continuous ground electrode (Figure 3.31.C). Finally, more E- Solder 3022 epoxy was manually applied from the backside of the backing layer to the edges of the 1 st matching layer, creating a pair of ground flaps, in order to strengthen the ground electrode connection on the front of the FL-IVUS probe (Figure 3.31.D). Figure 3.31. Connection of the 48 AWG micro-coaxial cable assembly to the flexible printed circuit. (A) Cores aligned and cold-soldered with E-Solder 3022 epoxy to their respective signal traces. All exposed outer conductors were also cold-soldered with each other using E-Solder 3022 epoxy. (B) All created connections were potted and protected with Epo-Tek 301 epoxy. (C) E-Solder 3022 epoxy applied to the backside of both branches of the flexible circuit as well as to the backside of the backing layer in order to create a continuous ground electrode. (D) E-Solder 3022 epoxy ground flaps created in order to strengthen the ground electrode connection on the front of the FL-IVUS probe. 152 3.7 Final Packaging and Connection to Verasonics Backshell Connector After cold-soldering the cable assembly, the soldering branches of the flexible printed circuit were carefully pulled inside the distal end of the 8F PTFE catheter. Epo-Tek 301 epoxy was circumferentially applied to join the 2.5-mm long piece of 8F polyimide tubing to the 0.55- m long piece of 8F PTFE tubing. Afterwards, a 15-μm-thick layer of Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN), was vapor deposited to serve as the 2 nd matching layer and to ensure complete insulation of the distal end of the FL-IVUS phased array probe (Figure 3.32.A). A size comparison of the distal end of the FL-IVUS phased array probe, the array acoustic stack–backing layer–flexible circuit sub-assembly, and a dime coin is shown in Figure 3.32.B. Figure 3.32. Final packaging of the FL-IVUS phased array probe. (A) Finished distal end with 15-μm layer of vapor deposited Parylene C to serve as the 2 nd matching layer and to ensure complete insulation. (B) Size comparison of the finished distal end of the FL-IVUS phased array probe, the array acoustic stack–backing layer–flexible circuit sub-assembly, and a dime coin. The proximal end of the cable assembly was then manually cold-soldered with E-Solder 3022 to a Verasonics micro-coax termination board (Figure 3.33.A). All the connections were 153 potted with Epo-Tek 301 for protection, leaving exposed just the top surface of the resulting E- Solder 3022 electrodes for posterior poling and testing. The bundle of 32 micro-coaxial cables was bonded with Epo-Tek 301 to a black PVC cable-strain relief boot (Amphenol Corporation, Wallingford, CT) and finished with a Verasonics M14-A2-70 Hex cable-strain relief nut (Figure 3.33.B). Figure 3.33. Cable assembly termination of the proximal end. (A) 48 AWG micro-coaxial cables cold-soldered to a Verasonics micro-coax termination board. (B) Protection with a black PVC cable-strain relief boot and finish with a Verasonics M14-A2-70 Hex cable-strain relief nut. To connect the resulting FL-IVUS phased array probe (Figure 3.34.A) to the Verasonics Backshell Connector, the micro-coax termination board was connected to the Left Canister PCB (printed circuit board) in position 1L (Figure 3.34.B). Afterwards, the Left Canister PCB – termination board connected set, together with the Cannon DL-260 connector plug and locking handle, were fitted inside one backshell half (Figure 3.34.C) and finally the other backshell half was secured in place with the included set of screws to completely finish the FL-IVUS phased array probe (Figure 3.34.D). 154 Figure 3.34. Connecting the FL-IVUS phased array probe to the Verasonics Backshell Connector. (A) Probe finished with Verasonics micro-coax termination board. (B) Connection schematics of the termination board to the left canister printed circuit board. (C) Fitting Verasonics connection electronics inside one backshell half. (D) Completely finished and packaged FL-IVUS phased array probe. 3.8 Array Transducer Characterization Upon fabrication completion of the FL-IVUS phased array probe, the next step consisted in individually re-polarizing each one of the 32 piezoelectric PMN-PT elements in air at room 155 temperature under an electric field of 20 kV/cm for 3 minutes using a Spellman-Bertan Model 210-02R high voltage power supply (Spellman High Voltage Electronics Corporation, Hauppauge, NY). Afterwards, several standard non-imaging transducer tests were performed on the array probe to characterize its performance including electrical impedance, pulse-echo response, insertion loss, combined electrical and acoustical crosstalk, as well as single-element azimuthal one-way angular response or directivity 3.8.1 Electrical Impedance A flexible printed circuit (Figure 3.35.A) was designed to provide the 32 piezoelectric elements in an array acoustic stack (2-2 composite + 1 st matching layer) with probing pads where electrical impedance measurements could be performed. The flexible circuit features thirty-two 12- m-wide signal traces with a 25- m-wide pitch, divided into two groups of 16 signal traces each, and terminated into 95- m-diameter probing pads, being all these surrounded by a ring common ground electrode. The flexible circuit was fabricated by Cicor (Cicor Advanced Microelectronics, Ulm, Germany). Figure 3.35.B shows a physical sample of the flexible circuit consisting of a 12.5-μm-thick base dielectric polyimide layer (Kapton) (DuPont, Wilmington, DE) with the signal traces and probing pads on the front, and a common ground electrode on the back that connects to the ring common ground electrode on the front thru metalized vias. All signal traces, probing pads, and ground electrodes were patterned over a 2-μm-thick copper layer and finished with a flash of gold. Figure 3.35.C shows an array acoustic stack bonded to the flexible circuit following the methodology previously presented in section 3.6.3 Bonding the Array Acoustic Stack and Backing Layer to the Flexible Circuit. E-Solder 3022 conductive epoxy was added to electrically connect the front side of the 1 st matching layer to the ring 156 common ground electrode. Likewise, a piece of 25 AWG stranded wire was bonded to the ring common ground electrode with E-Solder 3022 in order to facilitate ground probing when performing the electrical impedance measurements. (A) (B) (C) Figure 3.35. Flexible printed circuit for electrical impedance measurements. (A) CAD schematics. (B) Physical sample. (C) Array acoustic stack and 25 AWG ground wire bonded to flexible printed circuit. 157 The electrical impedance of each individual array element was measured with an Agilent E4991A RF Impedance/Material Analyzer (Agilent Technologies, Santa Clara, CA) and both magnitude and phase angle were recorded over the frequency range of the array transducer pass- band. The simulated and measured electrical impedance magnitude and phase angle of a representative array element (element 17) are shown in Figures 3.36.A and 3.36.B, respectively. Element 17 was chosen as representative since its electrical impedance characteristics were closest to the average of all 32 elements. (A) (B) Figure 3.36. (A) Simulated and (B) measured electrical impedance magnitude and phase angle of representative array element 17. The simulated and measured results are summarized and compared in Table XVI. The measured results showed an electrical impedance magnitude of 118.9 ± 32.1 Ω at 30 MHz. The series (fr) and parallel (fa) resonant frequencies were 33.1 ± 2.8 MHz and 40.8 ± 3.1 MHz, respectively, yielding and electromechanical coupling coefficient (kt) of 0.62 ± 0.07, close to the reported value of 0.58 for bulk PMN-PT single crystal by CTS Corporation (CTS Corporation, Bolingbrook, IL). 158 Table XVI. Comparison of simulated and measured electrical impedance results. Parameter PZFlex Measured Electrical Impedance (Ω) 109.5 @ 30 MHz 118.9 ± 32.1 @ 30 MHz fr (MHz) 32.5 33.1 ± 2.8 fa (MHz) 42.1 40.8 ± 3.1 kt 0.67 0.62 ± 0.07 The uniformity of the measured values of electrical impedance magnitude and phase angle of all the array elements at 30 MHz is shown in Figure 3.37. As it can be noticed, there were only three open elements: 18, 21, and 23. Previously, these three elements presented a short circuit to ground, reason why the connections were opened by cutting the inner conductors cold- soldered to their respective electrodes on the Verasonics micro-coax termination board. The average and standard deviation of the array elements electrical impedance magnitude and phase angle were (118.9 ± 32.1 Ω) and (-51.4 ± 4.1°), respectively. Figure 3.37. Uniformity of electrical impedance magnitude and phase angle values for all the array elements. 159 3.8.2 Pulse-Echo Response and Insertion Loss The pulse-echo response of each individual array element was recorded to determine their effective center frequency, bandwidth, sensitivity, and pulse length. This test was performed by immersing the FL-IVUS phased array probe in a deionized water tank containing a polished quartz reflector as the target at the elevation natural focal distance of 5.0 mm. Each array element was pulsed with a 30-Volt, single-cycle sinusoid at the array center frequency. Received echo signals were amplified with a 29-dB gain and went through a bandpass filter from 20 MHz to 55 MHz. For each pulse-echo time domain signal, the -6 dB fractional bandwidth was determined using a Fast Fourier Transform (FFT), where the lower and upper bandwidth edges were determined by the frequencies at which the power spectrum was equal to -6 dB relative to the maximum value. The effective center frequency was taken as the middle point between the lower and upper limits of the -6 dB fractional bandwidth. Echo peak-to-peak amplitude was recorded for sensitivity and the -6 dB / -20 dB signal pulse lengths were determined by measuring the time between the first and last points, at which the signal was -6 dB / -20 dB relative to the maximum echo signal value, respectively. Insertion loss was measured by immersing the FL-IVUS phased array probe in a deionized water tank, exciting each element in the array with a 5-Vpp, 30-cycle sinusoidal tone- burst signal generated by a Tektronix AFG 3252 Dual Channel Arbitrary / Function Generator (Tektronix, Inc., Beaverton, OR) at the array center frequency, and receiving the reflected echo from a polished quartz reflector placed at the elevation natural focal distance. The receive power across a 50-Ω load was referenced to the source power delivered to a 50-Ω reference load and 160 expressed in decibels (Cannata et al., 2005). The measured values were afterwards corrected for loss due to attenuation in water (2.0 x 10 -4 dB/mm-MHz 2 ) (Lockwood et al., 1994) and reflection from the quartz target (1.9 dB). The simulated and measured pulse-echo response of a representative array element (element 22) are shown in Figures 3.38.A and 3.38.B, respectively. Element 22 was chosen as representative since its pulse-echo response characteristics were closest to the average of all 32 elements. (A) (B) Figure 3.38. (A) Simulated and (B) measured pulse-echo response of representative array element 22. The simulated and measured results are summarized and compared in Table XVII. The measured results showed an effective -6 dB center frequency (-6 dB Fc) of 28.9 ± 1.8 MHz, a -6 dB fractional bandwidth (-6 dB BW) of 36.4 ± 8.3%, a peak-to-peak sensitivity (Vpp) of 25.1 ± 8.9 mV, and a compensated insertion loss (IL) of 58.7 ± 3.6 dB. The overall decrease in sensitivity resulted mainly from the fact that all the array elements did not have their entire bottom surface perfectly connected to their respective flexible circuit signal electrodes, as it was the case of the PZFlex simulation model. In reality, each element in the array was sub-etched in 161 the elevation direction, resulting in 5-µm-wide sub-element kerfs, as advised by CTS Corporation to mitigate the risk of element electrodes peeling off the extremely narrow composite pillars due to resonance-induced strain; whereas the PZFlex simulation model utilized continuous elements in the elevation direction. Furthermore, the resulting Epo-Tek 301 epoxy thin bonding layer was not perfectly uniform after the clamping process. This can be verified by closely examining the cross-sectional image of the array acoustic stack shown in Figure 3.28.A and by observing the peak-to-peak sensitivity uniformity (Figure 3.39). The measured compensated insertion loss was higher than that reported for other linear arrays with similar construction or operating within the same frequency range; however, the element footprint and, consequently, the dimensions of the signal electrodes on the flexible printed circuit in this array were at least 3 times smaller, making it more difficult to ensure a perfect connection between the bottom surface of the array elements and their respective flexible circuit signal electrodes. For example, the forward-looking phased array developed by Stephens et al. (Stephens et al., 2009) featured 24 elements within an aperture of 1.2 mm by 1.58 mm, at a pitch of 65 μm, and a center frequency of 14 MHz; whereas the forward-looking phased array being presented included 8 more elements within an aperture of 0.8 mm by 1.0 mm, at a pitch of only 25 μm, and a center frequency of 28.9 MHz. Another forward-looking array operating at 30 MHz was developed by Ritter et al. (Ritter et al., 2002). Although that array featured more elements, 48, the aperture was 1.5 mm by 4.8 mm at a pitch of 100 μm to ease the burden of interconnecting to small array elements. Additionally, the 32 individual 48-AWG micro-coaxial cables used in the array under discussion were not shielded as a bundle, making them more susceptible to external noise and interference. A solution could be to assemble them together with binder tape, shield them with metallic braiding, and wrap them with additional 162 fluoropolymer tape (Stephens et al., 2008). This could not only decrease insertion loss but also improve sensitivity. Table XVII. Comparison of simulated and measured pulse-echo response results. Parameter PZFlex Measured -6 dB Fc (MHz) 27.1 ± 0.6 28.9 ± 1.8 -6 dB BW (%) 41.2 ± 6.3 36.4 ± 8.3 Vpp (mV) 33.6 ± 5.2 25.1 ± 8.9 -6 dB / -20 dB pulse length (nsec) 78 / 146 73 / 138 Compensated IL (dB) N/A 58.7 ± 3.6 The uniformity of the measured values of -6 dB center frequency, -6 dB fractional bandwidth, and peak-to-peak sensitivity of all the array elements is shown in Figure 3.39. Figure 3.39. Uniformity of -6 dB center frequency, -6 dB fractional bandwidth, and peak-to- peak sensitivity of all the array elements. 163 3.8.3 Electrical and Acoustical Crosstalk Combined electrical and acoustical crosstalk measurements were performed by immersing the FL-IVUS phased array probe in a deionized water tank. A Tektronix AFG 3252 Dual Channel Arbitrary / Function Generator (Tektronix, Inc., Beaverton, OR) was used to excite one element in the array with a 5-Vpp, 5-cycle sinusoidal tone-burst signal within a frequency range from 10 MHz to 50 MHz in steps of 2 MHz. The applied signal was measured with a LeCroy LC534 1-GHz Oscilloscope (LeCroy Corporation, Chestnut Ridge, NY) and served as a reference to the measured signals from four adjacent elements. These measurements were executed on element 16 with respective adjacent elements 15, 14, 13, and 12 as well as on element 8 with respective adjacent elements 7, 6, 5, and 4. Afterwards, the crosstalk for first adjacent elements, second adjacent elements, third adjacent elements, and fourth adjacent elements were averaged and plotted as relative amplitudes. The simulated and measured crosstalk values are shown in Figures 3.40.A and 3.40.B, respectively. (A) (B) Figure 3.40. (A) Simulated and (B) measured crosstalk values for four nearest neighboring elements in the array. 164 Maximum measured crosstalk values at the array center frequency were -26.5 dB for the first adjacent element, -30.7 dB for the second adjacent element, -34.7 dB for the third adjacent element, and -39.1 dB for the fourth adjacent element, which indicated satisfactory but not ideal element-to-element isolation according to the suggested maximum < -30 dB crosstalk design guideline for arrays with a linear configuration (Shung, 2006). The relatively high crosstalk value for the first adjacent element could be attributed to two main factors: 1. Acoustical cross- coupling between consecutive elements in the 2-2 array composite due to the narrow, 6-μm- wide, epoxy-filled kerfs. 2. Electrical cross-coupling between consecutive signal electrodes on the flexible printed circuit having a pitch of only 25 μm. As a comparison, Cummins et al. (Cummins et al., 2016) observed a maximum crosstalk value of -23.7 dB between consecutive elements on a 60-MHz array having 6-μm-wide, epoxy-filled kerfs. Likewise, Chiu et al. (Chiu et al., 2017) observed a maximum crosstalk value of -28 dB between consecutive elements on a 20-MHz array featuring a flexible printed circuit with a pitch of 37 μm between consecutive signal electrodes. Nonetheless, as it can be seen, the measured values of crosstalk were lower than their simulated counterparts. This mainly resulted from the fact that there were minuscule void spaces (cross-sectional area of 13 μm x 2 μm) defined by consecutive signal electrodes on the flexible printed circuit and the 2-2 composite, whereas in the FEA simulation model such void spaces were not taken into consideration. 3.8.4 Single-Element Directivity Response The one-way azimuthal directivity response was measured by rotating a representative array element around an axis along its length and center. The array element was excited with a 50-Vpp, 5-cycle sinusoidal tone-burst signal generated by a Tektronix AFG 3252 Dual Channel 165 Arbitrary / Function Generator connected to an Amplifier Research 50W1000B solid-state power amplifier (Amplifier Research, Inc., Souderton, PA). A needle hydrophone HGL-0085 (Onda Corporation, Sunnyvale, CA), placed at the natural elevation focus and connected to a LeCroy WaveRunner 104MXi 1-GHz Oscilloscope, was used to acquire the amplitude of the time- domain response at discrete angular positions (Cannata et al., 2005). The Field II program (Jensen, 1996) was used to simulate the directivity of a single element in the array and to estimate the effective element width by matching the obtained simulated directivity curve to the measured values (Cannata et al., 2006). The simulated and measured one-way azimuthal directivity response of representative array element 22 are shown in Figure 3.41. Figure 3.41. Simulated and measured one-way azimuthal directivity response of representative array element 22. The measured -6 dB directivity was approximately ±20°. The higher-than-desired measured maximum crosstalk explains why the effective element width (36 μm) was approximately 45% larger than the actual array pitch of 25 μm. A similar result was reported by 166 Cannata et al. of an effective element width 40% larger than the actual array pitch for a 35-MHz linear array with higher-than-desired maximum crosstalk < -24 dB and -6 dB directivity of ±20° (Cannata et al., 2006). Likewise, Chiu et al. reported a -6 dB directivity of approximately ±20° for a 20-MHz phased array with higher-than-desired maximum crosstalk < -28 dB (Chiu et al., 2017). Table XVIII comprehensively summarizes and compares the simulated and measured results for the array transducer characterization testing. Table XVIII. Comparison summary of simulated and measured characterization results. Parameter PZFlex Measured Electrical impedance (Ω) 109.5 @ 30 MHz 118.9 ± 32.1 @ 30 MHz Series resonant frequency, fr (MHz) 32.5 33.1 ± 2.8 Parallel resonant frequency, fa (MHz) 42.1 40.8 ± 3.1 Electromechanical coupling coefficient, kt 0.67 0.62 ± 0.07 -6 dB center frequency (MHz) 27.1 ± 0.6 28.9 ± 1.8 -6 dB fractional bandwidth (%) 41.2 ± 6.3 36.4 ± 8.3 Peak-to-peak sensitivity (mV) 33.6 ± 5.2 25.1 ± 8.9 -6 dB / -20 dB pulse length (nsec) 78 / 146 73 / 138 Compensated insertion loss (dB) N/A 58.7 ± 3.6 Electrical and acoustical crosstalk (dB) < -18.5 < -26.5 Single-element directivity response 1 ± 20° ± 20° 1. Simulated value obtained with Field II program. 3.9 Array Transducer Imaging The ultimate performance indicators of the FL-IVUS phased array probe were determined by its imaging capability of three different kinds of targets using a Verasonics Vantage 128 System. Imaging was performed using multiple ray lines in a phased array configuration, with each ray line constituting a separate beamformed transmit operation, followed by a respective beamformed receive operation. Radio frequency (RF) data was acquired 167 using the Verasonics “4/3 Sampling” mode (Verasonics, 2017), with the transmit frequency set to 31.25 MHz. After all of the individual ray lines were acquired, the image data was reconstructed to yield the phased array images. No apodization was applied to the array transducer aperture in any of the imaging experiments. 3.9.1 Polished Quartz Reflector Imaging The first target (Figure 3.42) was a polished quartz reflector immersed in a tank with deionized water and placed at three different depths: 1. Approximately half the elevation natural focal distance, 2. Elevation natural focal distance, and 3. Twice the elevation natural focal distance. For these three cases, respectively, the transmit focus was set at a depth of 3 mm with a transmit pulse of 22 Vpp, a depth of 5 mm with a transmit pulse of 36 Vpp, and a depth of 10 mm with a transmit pulse of 29 Vpp. The acquired images were post-processed with a scaling gain of 40 dB. Figure 3.42. Polished quartz reflector for imaging experiments of the FL-IVUS phased array probe. The polished quartz reflector was immersed in a tank with deionized water. 168 The acquired images of the polished quartz reflector are shown in Figure 3.43. As it can be seen, the array transducer was capable of imaging such target up to a depth of twice the elevation natural focal distance. As expected, when the quartz reflector was placed in the near field at approximately half the elevation natural focal distance (Figure 3.43.A) the field of view was reduced and cluttering could be seen all along the imaged surface. When the quartz reflector was placed at the elevation natural focal distance (Figure 3.43.B) the field of view increased and cluttering was reduced in the central region of the imaged surface. When the quartz reflector was placed in the far field at twice the elevation natural focal distance (Figure 3.43.C) the field of view continued to increase and cluttering was reduced all along the imaged surface. (A) (B) (C) Figure 3.43. Imaging of polished quartz reflector. (A) Target located in the near field at half the elevation natural focal distance. (B) Target located at the elevation natural focal distance. (C) Target located in the far field at twice the elevation natural focal distance. 3.9.2 Custom-Made Fine-Wire Phantom Imaging The second target (Figure 3.44) was a custom-made fine-wire phantom composed of three evenly spaced 20-μm-diameter tungsten wires (California Fine Wire Company, Grover Beach, CA) with the finality of determining the axial and lateral spatial resolutions of the array transducer. The axial separation between wires was around 1.7 mm and the azimuth separation 169 was around 0.5 mm. The custom-made fine-wire phantom was immersed in a tank with deionized water. At a center frequency of 30 MHz, the sound wavelength in the transmitting medium is 50 μm; therefore, the wires could be assumed as three point targets. The transmit focus was set at a depth of 5 mm with a transmit pulse of 40 Vpp. The acquired image was post- processed with a scaling gain of 10 dB and minimal thresholding was used in order to highlight the wire targets. Figure 3.44. Custom-made fine-wire phantom for imaging experiments of the FL-IVUS phased array probe. The phantom was immersed in a tank with deionized water. The acquired image of the custom-made fine-wire phantom is shown in Figure 3.45.A. As it can be seen, the three wires were clearly visible with reasonable image quality, especially the middle one located at the elevation natural focal distance. Faint artifacts were observed mostly to the left and right sides of the three wires due to the presence of sidelobes in the ultrasound beam. Implementing apodization could have reduced the amplitude of such sidelobes; however, this would have increased the main lobe width, consequently degrading spatial resolutions (Frazier and O’Brien, 1998). Plots of the axial and lateral line spread functions for the middle wire are shown in Figures 3.45.B and 3.45.C, respectively. The measured full-width 170 half-maximum (FWHM) spatial resolutions were around 65 μm and 215 μm in the axial and lateral directions, respectively. These measurements correlated well with the theoretical axial (54.75 μm) and lateral (250 μm) spatial resolutions, predicted by equations (1.20) to (1.22). Figure 3.45. Imaging of custom-made fine-wire phantom. (A) Acquired image. (B) Plot of the axial line spread function for the middle wire. (C) Plot of the lateral line spread function for the middle wire. 3.9.3 Custom-Made Porcine Carotid Artery Phantom Imaging The third and last target (Figure 3.46) was composed of tissue-mimicking urethane rubber (α = 0.5 dB/cm-MHz, c = 1,450 m/s) (Supertech, Inc., Elkhart, IN) into which a piece of porcine carotid artery having a maximum diameter of approximately 3 mm was inserted with the finality of demonstrating the capability of the array transducer to image the intraluminal path and structures ahead of the catheter. Positioning the array transducer at different depths inside such artery allowed imaging several sections having different diameters with and without blockages 171 artificially created with small fragments of urethane rubber. The custom-made porcine carotid artery phantom was immersed in a tank with deionized water. The transmit focus was set at a depth of 8 mm with a transmit pulse of 50 Vpp. The acquired images were post-processed with a scaling gain of 30 dB. Figure 3.46. Custom-made porcine carotid artery phantom for imaging experiments of the FL- IVUS phased array probe. The phantom was immersed in a tank with deionized water. The acquired images of the custom-made porcine carotid artery phantom are shown in Figures 3.47 to 3.49. Figure 3.47 shows the widest section of the artery having a diameter of approximately 3 mm under increasing levels of occlusion. As it can be seen, the arterial walls were clearly visible up to a depth ranging from 8 mm to 10 mm ahead of the array transducer catheter. Figure 3.47.A shows an occlusion-free arterial lumen. Figure 3.47.B shows an approximately half- occluded arterial lumen. Figure 3.47.C shows an increased level of occlusion, which left only a narrow free passage on the bottom right section of the arterial lumen. Figure 3.47.D shows a completely occluded arterial lumen. 172 Figure 3.47. Imaging of custom-made porcine carotid artery phantom – widest section, 3-mm diameter. (A) Occlusion-free arterial lumen. (B) Half-occluded arterial lumen. (C) Increased level of occlusion, only a narrow free passage on the bottom right section of the arterial lumen. (D) Completely occluded arterial lumen. Figure 3.48 shows a transition section of the artery at which the diameter reduced from 3 mm to approximately 2.5 mm. Again, the arterial walls were clearly visible up to a depth of 8 mm ahead of the array transducer catheter. Figure 3.48.A shows an occlusion-free arterial lumen. Figure 3.48.B shows an occlusion that left only a narrow free passage on the bottom left section of the arterial lumen. 173 Figure 3.48. Imaging of custom-made porcine carotid artery phantom – transition section, from 3-mm to 2.5-mm diameter. (A) Occlusion-free arterial lumen. (B) Occlusion leaving only a narrow free passage on the bottom left section of the arterial lumen. Figure 3.49 shows another transition section of the artery at which the diameter reduced from 2.5 mm to approximately 1 mm. This section of the artery was the deepest one that could be imaged due to the size of the array transducer catheter. Once again, the arterial walls were clearly visible up to a depth of 8 mm ahead of the array transducer catheter. Figure 3.49.A shows an occlusion-free arterial lumen. Figure 3.49.B shows an occlusion that left only a narrow free passage on the bottom left section of the arterial lumen. Figure 3.49. Imaging of custom-made porcine carotid artery phantom – transition section, from 2.5-mm to 1-mm diameter. (A) Occlusion-free arterial lumen. (B) Occlusion leaving only a narrow free passage on the bottom left section of the arterial lumen. 174 CHAPTER 4 SUMMARY 4.1 Miniature Linear Phased Array for Colorectal Robotic Surgery 4.1.1 Array Development Summary The design, simulation modeling, and fabrication procedure of a 15-MHz, SL-ERUS linear phased array have been described. The array is intended to be integrated into a robotic surgery probe for the real-time intraoperative fusion of ERUS and endomicroscopy to assist in the assessment of CRC surgical margins during TEM procedures. The array features 64 piezoelectric elements separated by non-conductive epoxy kerfs, yielding a total active aperture of 3.2 mm in the azimuth direction and 1.8 mm in the elevation direction, with an elevation natural focal depth of 8.1 mm. The array includes non-conductive epoxy backing and two front matching layers. The optimized simulated performance confirmed that the array provides enough penetration (≥ 8.1 mm), peak-to-peak sensitivity (≥ 100 mV), and spatial resolution (80.25 μm axially / 450 μm laterally) for the overall structure detection of CRC tissue. The array fabrication procedure confirmed that the most challenging aspect of fabricating this type of ultrasound probe is the connection of the miniature high-frequency ultrasound array to the imaging system via an electrical interconnect solution consisting of a custom flexible circuit, micro-coaxial cabling, and a specialized system connector; where correctly aligning and connecting the array elements to the flexible circuit, as well as soldering the micro-coaxial cables to their respective flexible circuit electrodes proved to be the most critical steps of the whole 175 fabrication process. The array was fitted inside of a 1.5-m long 10F catheter having an external diameter of only 3.3 mm. 4.1.2 Array Performance Summary Several standard non-imaging transducer tests were performed on the array probe to characterize its performance including electrical impedance, pulse-echo response, insertion loss, combined electrical and acoustical crosstalk, as well as single-element azimuthal one-way angular response or directivity. Additionally, the ultimate performance indicators of the array probe were determined by its imaging capability of a polished quartz reflector, a fine-wire phantom, and tissue-mimicking phantoms using a Verasonics Vantage 128 System. Imaging results of the tissue-mimicking phantoms successfully demonstrated the capability of the SL- ERUS phased array probe to detect solid structures and cysts present in tissue; therefore confirming its suitability for colorectal tissue characterization during TEM procedures. Table XIX below comprehensively summarizes the SL-ERUS linear phased array design parameters and the comparison of simulated and measured performance results. Table XIX. SL-ERUS linear phased array design parameters and comparison summary of simulated and measured performance results. Design Parameters Design center frequency 15 MHz Number of elements 64 Composite configuration 2-2, linear Pitch (0.5 water) 50 m Element width 37 m Kerf width 13 m Elevation aperture 1.8 mm Azimuth aperture 3.2 mm Elevation focal distance (natural) 8.1 mm Piezoelectric material PMN-PT 53 m thick 176 Kerf filler material Non-conductive epoxy (Epo-Tek 301) 53 m thick Backing material Non-conductive epoxy (Epo-Tek 301) with 43.2wt% liquid plasticizer (LP-3) 2 mm thick 1 st Matching layer material Conductive epoxy (2,3- m silver epoxy) 45 m thick 2 nd Matching layer material Vapor deposited polymer (Parylene C) 30 m thick Parameter PZFlex Measured Electrical impedance ( ) 123.8 @ 15 MHz 131.3 ± 27.1 @ 15 MHz Series resonant frequency, fr (MHz) 15.8 17.6 ± 1.2 Parallel resonant frequency, fa (MHz) 19.9 20.7 ± 1.6 Electromechanical coupling coefficient, kt 0.64 0.57 ± 0.05 -6 dB center frequency (MHz) 14.1 ± 0.5 17.7 ± 1.2 -6 dB fractional bandwidth (%) 56.7 ± 2.1 52.2 ± 9.8 Peak-to-peak sensitivity (mV) 238.9 ± 98.7 200.1 ± 102.7 -6 dB / -20 dB pulse length (nsec) 137 / 211 107 / 193 Compensated insertion loss (dB) N/A 49.8 ± 3.9 Electrical and acoustical crosstalk (dB) < -20.8 < -30.5 Single-element directivity response 1 ± 22° ± 22° FWHM axial resolution (μm) 80.25 90 FWHM lateral resolution (μm) 450 420 1. Simulated value obtained with Field II program. 4.2 Forward-Looking Intravascular Imaging Catheter 4.2.1 Array Development Summary The design, simulation modeling, and fabrication procedure of a 30-MHz, FL-IVUS linear phased array have been described. The array is intended to be used for intravascular imaging and intraluminal vessel navigation of guide wires during PAD angioplasty procedures. The array features 32 piezoelectric elements separated by non-conductive epoxy kerfs, yielding a total active aperture of 0.8 mm in the azimuth direction and 1.0 mm in the elevation direction, with an elevation natural focal depth of 5.0 mm. The array includes non-conductive epoxy 177 backing and two front matching layers. The optimized simulated performance confirmed that the array provides enough depth of view (≥ 5.0 mm), peak-to-peak sensitivity (≥ 25 mV), and spatial resolution (54.75 μm axially / 250 μm laterally) for the overall detection of structures lying ahead of the catheter. The array fabrication procedure confirmed that the most challenging aspect of fabricating this type of ultrasound probe is the connection of the miniature high-frequency ultrasound array to the imaging system via an electrical interconnect solution consisting of a custom flexible circuit, micro-coaxial cabling, and a specialized system connector; where correctly aligning and connecting the array elements to the flexible circuit, as well as soldering the micro-coaxial cables to their respective flexible circuit electrodes proved to be the most critical steps of the whole fabrication process. The array was fitted inside of a 0.6-m long 8F catheter having an external diameter of only 2.6 mm. 4.2.2 Array Performance Summary Several standard non-imaging transducer tests were performed on the array probe to characterize its performance including electrical impedance, pulse-echo response, insertion loss, combined electrical and acoustical crosstalk, as well as single-element azimuthal one-way angular response or directivity. Additionally, the ultimate performance indicators of the array probe were determined by its imaging capability of a polished quartz reflector, a fine-wire phantom, and a porcine carotid artery phantom using a Verasonics Vantage 128 System. Imaging results of the porcine carotid artery phantom successfully demonstrated the capability of the FL- IVUS phased array probe to visualize the intraluminal path and structures ahead of the catheter; 178 therefore confirming its suitability for the navigation of catheters and guide wires during peripheral angioplasty procedures of highly stenosed lesions. Table XX below comprehensively summarizes the FL-IVUS linear phased array design parameters and the comparison of simulated and measured performance results. Table XX. FL-IVUS linear phased array design parameters and comparison summary of simulated and measured performance results. Design Parameters Design center frequency 30 MHz Number of elements 32 Composite configuration 2-2, linear Pitch (0.5 water) 25 m Element width 19 m Kerf width 6 m Elevation aperture 1 mm Azimuth aperture 0.8 mm Elevation focal distance (natural) 5 mm Piezoelectric material PMN-PT 27 m thick Kerf filler material Non-conductive epoxy (Epo-Tek 301) 27 m thick Backing material Non-conductive epoxy (Epo-Tek 301) with 43.2wt% liquid plasticizer (LP-3) 2 mm thick 1 st Matching layer material Conductive epoxy (2,3- m silver epoxy) 25 m thick 2 nd Matching layer material Vapor deposited polymer (Parylene C) 15 m thick Parameter PZFlex Measured Electrical impedance (Ω) 109.5 @ 30 MHz 118.9 ± 32.1 @ 30 MHz Series resonant frequency, fr (MHz) 32.5 33.1 ± 2.8 Parallel resonant frequency, fa (MHz) 42.1 40.8 ± 3.1 Electromechanical coupling coefficient, kt 0.67 0.62 ± 0.07 -6 dB center frequency (MHz) 27.1 ± 0.6 28.9 ± 1.8 -6 dB fractional bandwidth (%) 41.2 ± 6.3 36.4 ± 8.3 Peak-to-peak sensitivity (mV) 33.6 ± 5.2 25.1 ± 8.9 -6 dB / -20 dB pulse length (nsec) 78 / 146 73 / 138 Compensated insertion loss (dB) N/A 58.7 ± 3.6 Electrical and acoustical crosstalk (dB) < -18.5 < -26.5 Single-element directivity response 1 ± 20° ± 20° 179 FWHM axial resolution (μm) 54.75 65 FWHM lateral resolution (μm) 250 215 1. Simulated value obtained with Field II program. 180 Bibliography American Cancer Society. (2015). What are the key statistics about breast cancer? Retrieved from: http://www.cancer.org/cancer/breastcancer/detailedguide/breast-cancer-key-statistics American Cancer Society. (2015). What are the key statistics about colorectal cancer? Retrieved from: http://www.cancer.org/cancer/colonandrectumcancer/detailedguide/colorectal-cancer-key- statistics American Cancer Society. (2015). What are the key statistics about lung cancer? Retrieved from: http://www.cancer.org/cancer/lungcancer-non-smallcell/detailedguide/non-small-cell-lung- cancer-key-statistics American Cancer Society. (2015). What are the key statistics about prostate cancer? Retrieved from: http://www.cancer.org/cancer/prostatecancer/detailedguide/prostate-cancer-key-statistics American Heart Association. (2015). Peripheral Artery Disease. Retrieved from: http://watchlearnlive.heart.org/CVML_Player.php?moduleSelect=periad American Institute of Ultrasound in Medicine & National Electrical Manufacturers Association. (2004). NEMA UD 3-2004, Standard for Real-Time Display of Thermal and Mechanical Acoustic Output Indices on Diagnostic Ultrasound Equipment, Revision 2. Analog Devices. (1999). Ultrasound Analog Electronics Primer. Retrieved from: http://www.analog.com/library/analogDialogue/archives/33-05/ultrasound/index.html Blausen Medical. (2015). Colon Cancer. Retrieved from: http://blausen.com/?Topic=1654 Boguszewski, A., Torey, J., Pai, R., Kamalakannan, D., Jefic, D., & Davis, T. (2010). Intraluminal recanalization of SFA CTOs. Endovascular Today, 33(8), 33-38. Cady, W. G. (1964). Piezoelectricity: An introduction to the theory and applications of electromechanical phenomena in crystals. New York, NY. Dover Publications. Cancer Research United Kingdom. (2015). Bowel Cancer (Colorectal Cancer). Retrieved from: http://www.cancerresearchuk.org/about-cancer/type/bowel-cancer/ Cannata, J. M., Ritter, T. A., Chen, W. H., Silverman, R. H., & Shung, K. K. (2003). Design of efficient, broadband single-element (20-80 MHz) ultrasonic transducers for medical imaging applications. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 50(11), 1548-1557. Cannata, J. M., Williams, J. A., & Shung, K. K. (2005). A Kerfless 30 MHz Linear Ultrasonic Array. IEEE Ultrasonics Symposium, 109-112. 181 Cannata, J. M., Williams, J. A., Zhou, Q., Ritter, T. A., & Shung, K. K. (2006). Development of a 35-MHz Piezo-Composite Ultrasound Array for Medical Imaging. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 53(1), 224-236. Centers for Disease Control and Prevention. (2015). Colorectal Cancer Statistics. Retrieved from: http://www.cdc.gov/cancer/colorectal/statistics/ Chaubal, N., Dighe, M., Shah, M., & Chaubal, J. (2006). Sonography of the Gastrointestinal Tract. J. Ultrasound Med., 25(1), 87-97. Chen, R., Cabrera-Munoz, N. E., Lam, K. H., Hsu, H. S., Zheng, F., Zhou, Q., & Shung, K. K. (2014). PMN-PT Single-Crystal High-Frequency Kerfless Phased Array. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 61(6), 1033-1041. Chen, Y., Lam, K. H., Zhou, D., Cheng, W. F., Dai, J. Y., Luo, H. S., & Chan, H. L. W. (2013). High-frequency PMN-PT single crystal focusing transducer fabricated by a mechanical dimpling technique. Ultrasonics, 53(2), 345-349. Chen, Y. C., Lin, Y. H., Wang, S. H., Lin, S. P., Shung, K. K., & Wu, C. C. (2014). Monitoring tissue inflammation and responses to drug treatments in early stages of mice bone fracture using 50 MHz ultrasound. Ultrasonics, 54(1), 177-186. Chiu, C. T., Kang, B. J., Eliahoo, P., Abraham, T., & Shung, K. K. (2017). Fabrication and Characterization of a 20-MHz Microlinear Phased-Array Transducer for Intervention Guidance. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 64(8), 1261-1268. Cobbold, R. S. C. (2007). Foundations of Biomedical Ultrasound. New York, NY. Oxford University Press. Cummins, T., Eliahoo, P., & Shung, K. K. (2016). High-Frequency Ultrasound Array Designed for Ultrasound-Guided Breast Biopsy. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 63(6), 817-827. De Graaf, E. J. R., Doornebosch, P. G., Stassen, L. P. S., Debets, J. M. H., Tetteroo, G. W. M., & Hop, W. C. J. (2002). Transanal endoscopic microsurgery for rectal cancer. European Journal of Cancer, 38(7), 904-910. Demartines, N., Von Flue, M. O., & Harder, F. H. (2001). Transanal endoscopic microsurgical excision of rectal tumors: indications and results. World J. Surg., 25(7), 870-875. Desilets, C. S., Fraser, J. D., & Kino, G. S. (1978). The design of efficient broad-band piezoelectric transducers. IEEE Trans. Sonics Ultrason., 25(3), 115-125. Dwyer, G., Giataganas, P., Pratt, P., Hughes, M., & Yang, G. Z. (2015). A Miniaturized Robotic Probe for Real-Time Intraoperative Fusion of Ultrasound and Endomicroscopy. IEEE International Conference on Robotics and Automation (ICRA), 1196-1201. 182 Edelman, B. R., & Weiser, M. R. (2008). Endorectal Ultrasound: Its Role in the Diagnosis and Treatment of Rectal Cancer. Clin. Colon Rectal Surg., 21(3), 167-177. European School of Oncology. (2011). The History of Colorectal Cancer Surgery. Retrieved from: http://www.slideshare.net/ESOSLIDES/mco-2011-slide-1 Evans, A., Whelehan, P., Thomson, K., McLean, D., Brauer, K., Purdie, C., Jordan, L., Baker, L., & Thompson, A. (2010). Quantitative shear wave ultrasound elastography: Initial experience in solid breast masses. Breast Cancer Research, 12(6), 1-11. Falow, O., Baddour, R. E., Nathanael, G., Czarnota, G. J., & Kumaradas, J. C. (2008). A study of high frequency ultrasound scattering from non-nucleated biological specimens. J. Acoust. Soc. Am., 124(5), 278-283. Finstad, M. (2008). Basics of Flex Circuit Design. Minco Products. Retrieved from: http://sccontent.minco.com/resourcecenter/instructionstechnicalspecs Food and Drug Administration. (1997). Information for Manufacturers Seeking Marketing Clearance of Diagnostic Ultrasound Systems and Transducers. Center for Devices and Radiological Health. Food and Drug Administration. Foster, F. S., Zhang, M. Y., Zhou, Y. Q., Liu, G., Mehi, J., Cherin, E., Harasiewicz, K. A., Starkoski, B. G., Zan, L., Knapik, D. A., & Adamson, S. L. (2002). A new ultrasound instrument for in vivo microimaging of mice. Ultrasound Med. Biol., 28(9), 1165-1172. Frazier, C. H. and O’Brien, W. D. (1998). Synthetic aperture techniques with a virtual source element. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 45(1), 196-207. Goss, S., Johnston, R., & Dunn, F. (1978). Comprehensive compilation of empirical ultrasonic properties of mammalian tissues. J. Acoust. Soc. Am., 64(2), 423-457. Haifeng, W., Ritter, T., Cao, W., & Shung, K. K. (2001). High Frequency Properties of Passive Materials for Ultrasonic Transducers. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 48(1), 78-84. Hewick, S. A., Fairhead, A. C., Culy, J. C., & Atta, H. R. (2004). A comparison of 10 MHz and 20 MHz ultrasound probes in imaging the eye and orbit. Br. J. Ophtalmol., 88(4), 551-555. Hu, C. H., Zhang, L., Cannata, J. M., Yen, J., & Shung, K. K. (2011). Development of a 64 channel ultrasonic high frequency linear array imaging system. Ultrasonics, 51(8), 953-959. Ignatov, V., Kolev, N., Tonev, A., Shterev, S., Encheva, E., Kirilova, T., Teneva, T., & Ivanov, K. (2014). Diagnostic Modalities in Colorectal Cancer – Endoscopy, CT and PET Scanning, Magnetic Resonance Imaging (MRI), Endoluminal Ultrasound and Intraoperative Ultrasound. InTech Open Science Open Minds, 29-51. 183 International Electrotechnical Commission. (2007). IEC 60601-2-37, Medical electrical equipment – Part 2-37: Particular requirements for the basic safety and essential performance of ultrasonic medical diagnostic and monitoring equipment. Interview with Karen Woo, M.D. Endovascular Surgeon. Keck Medical Center of the University of Southern California. February 13, 2015. Interview with Stephen Eric Ryan, M.D. Former Vice-President, Clinical and Medical Affairs, Volcano Corporation. University of Southern California, University Park Campus. January 29, 2015. Interview with Sukgu M. Han, M.D. Endovascular Surgeon. Keck Medical Center of the University of Southern California. February 4, 2015. IPC-Association Connecting Electronics Industries. (2016). IPC-2223D, Sectional Design Standard for Flexible/Rigid-Flexible Printed Boards. Jensen, J. A. (1996). FIELD: A Program for Simulating Ultrasound Systems. Med. Biol. Eng. Comp., 34(1), 351-353. Kim, H. H. (2010). Array Transducers for High-Frequency Ultrasound Imaging. (Doctoral dissertation). Retrieved from: http://digitallibrary.usc.edu/cdm/compoundobject/collection/p1579 9coll127/id/310531/rec/4 Kim, H. H., Chang, J. H., Cannata, J. M., & Shung, K. K. (2007). Design of 20 MHz Convex Array Transducers for High Frequency Ophthalmic Imaging. IEEE Ultrasonics Symposium, 88- 91. Kim, H. H., Cannata, J. M., Williams, J. A., Chang, J. H., & Shung, K. K. (2009). Fabrication of 20 MHz Convex Array Transducers for High Frequency Ophthalmic Imaging. IEEE International Ultrasonics Symposium, 1130-1133. Kino, G. S. (1987). Acoustic waves: Devices, imaging, and analog signal processing. Englewood Cliffs, NJ. Prentice Hall. Kinsler, L. E., Frey, A. R., Coppens, A. B., & Sanders, J. V. (1999). Fundamentals of Acoustics. 4 th Edition. New York, NY. John Wiley & Sons. Krimholtz, R., Leedom, D. A., & Matthaei, G. L. (1970). New equivalent circuit for elementary piezoelectric transducers. Electron. Lett., 6(1), 398-399. Krishnan, P., Baber, U., Purushothaman, K., Purushothaman, M., Wiley, J., Tarricone, A., Mehran, R., Kovacic, J., Kim, M., Moreno, P., Kini, A., Dangas, G., & Sharma, S. (2012). Histopathologic evidence of adventitial cuts predicts restenosis after directional atherectomy of lower extremity peripheral arterial disease: Results from a randomized, open label, investigator- initiated trial comparing intravascular ultrasound-guided atherectomy to angiography guided 184 atherectomy in peripheral vascular interventions for TASC’s A, B Lesions (Utopia) Pilot Study. JACC, 59(13), E2083. Lethiecq, M., Feuillard, G., Ratsimandresy, L., Nguyen-Dinh, A., Pardo, L., Ricote, J., Andersen, B., & Millar, C. (1994). Miniature high frequency array transducers based on new fine grain ceramics. IEEE Ultrasonics Symposium, 2(1), 1009-1013. Liu, C., Djuth, F. T., Zhou, Q., & Shung, K. K. (2013). Micromachining techniques in developing high-frequency piezoelectric composite ultrasonic array transducers. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 60(12), 2615-2625. Lockwood, G. R., & Foster, F. S. (1995). Design of Sparse Array Imaging Systems. IEEE Ultrasonics Symposium, 2(1), 1237-1243. Lockwood, G. R., Li, P. C., O’Donnell, M., & Foster, F. S. (1996). Optimizing the Radiation Pattern of Sparse Periodic Linear Arrays. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 43(1), 7-14. Lockwood, G. R., Turnbull, D. H., & Foster F. S. (1994). Fabrication of High Frequency Spherically Shaped Ceramic Transducers. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 41(2), 231-235. Marsal, K. (2005). The output display standard: has it missed its target? ISUOG Ultrasound Obstet. Gynecol., 25(3), 211-214. Maslak, S. (1985). Computed Sonography. In Sanders, R., & Hill, M., editors. Ultrasound Annual. New York, NY. Raven Press. McCloud, J. M., Waymont, N., Pahwa, N., Varghese, P., Richards, C., Jameson, J. S., & Scott, A. N. D. (2006). Factors predicting early recurrence after transanal endoscopic microsurgery excision for rectal adenoma. Colorectal Disease, 8(1), 581-585. Medical Billing and Coding Online. (2015). Colorectal Cancer Screening ICD9 and CPT Codes. Retrieved from: http://www.medicalbillingcodings.org/2014/10/Colorectal-cancer-screening- icd9cpt-codes.html Meuwly, J. Y., & Gudinchet, F. (2004). Sonography of the thoracic and abdominal walls. J. Clin. Ultrasound, 32(9), 500-510. Nelson, T. R., Fowlkes, J. B., Abramowicz, J. S., & Church, C. C. (2009). Ultrasound Biosafety Considerations for the Practicing Sonographer and Sonologist. AIUM J. Ultrasound Med., 28(2), 139-150. Newnham, R. E., Skinner, D. P., & Cross, L. E. (1978). Connectivity and piezoelectric- pyroelectric composites. Mater. Res. Bull., 13(5), 525-536. 185 Norgren, L., Hiatt, W. R., Dormandy, J. A., Nehler, M. R., Harris, K. A., & Fowkes, F. G. R. (2007). Inter-Society Consensus for the Management of Peripheral Arterial Disease (TASC II). Journal of Vascular Surgery, (45)1, S5A-S67A. Olympus. (2006). Ultrasonic Transducers Technical Notes. Retrieved from: https://www.olympus-ims.com/data/File/panametrics/UT-technotes.en.pdf Ophir, J., Maklad, N. F., & Bigelow, R. H. (1982). Ultrasonic attenuation measurements of in vivo human muscle. Ultrasonic Imaging, 4(3), 290-295. Panza, J. A., Petrone, R. K., Fananapazir, L., & Maron, B. J. (1992). Utility of continuous wave Doppler echocardiography in the noninvasive assessment of left ventricular outflow tract pressure gradient in patients with hypertrophic cardiomyopathy. J. Am. Coll. Cardiol., 19(1), 91- 99. Penninck, D., & d’Anjou, M. A. (2015). Atlas of Small Animal Ultrasonography. 2 nd Edition. United States of America. Wiley-Blackwell. Pozar, D. M. (2012). Microwave Engineering. 4 th Edition. United States of America. John Wiley & Sons, Inc. Pratt, P., Di Marco, A., Payne, C. J., Darzi, A., & Yang, G. Z. (2012). Intraoperative ultrasound guidance for transanal endoscopic microsurgery. Medical Image Computing and Computer- Assisted Interventions MIC-CAI, 7510(1), 463-470. Prince, J. L., & Links, J. (2014). Medical Imaging Signals and Systems. 2 nd Edition. United States of America. Pearson Prentice Hall. Ritter, T. A., Shrout, T. R., Tutwiler, R., & Shung, K. K. (2002). A 30-MHz Piezo-Composite Ultrasound Array for Medical Imaging Applications. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 49(2), 217-230. Robinson, D., Wilson, L., & Kossoff, G. (1981). Shadowing and enhancement in ultrasonic echograms by reflection and refraction. J. Clin. Ultrasound, 9(4), 181-188. Savakas, H. P., Klicker, K. A., & Newnham, R. E. (1981). PZT-epoxy piezoelectric transducers: A simplified fabrication procedure. Materials Res. Bull., 16(1), 677-680. Shung, K. K. (2006). Diagnostic Ultrasound: Imaging and Blood Flow Measurements. Boca Raton, FL. CRC Press. Siemens Healthcare. (2015). ACUSON S3000 TM Ultrasound System. Retrieved from: http://www.healthcare.siemens.com/ultrasound/obgyn/acuson-s3000 186 Silverman, R. H., Rondeau, M. J., Lizzi, F. L., & Coleman, D. J. (1995). Three-dimensional high-frequency ultrasonic parameter imaging of anterior segment pathology. Ophthalmology, 102(5), 837-843. Starck, M., Bohe, M., Simanaitis, M., & Valentin, L. (2003). Rectal endosonography can distinguish benign rectal lesions from invasive early rectal cancers. Colorectal Dis., 5(3), 246- 250. Stephens, D. N., Cannata, J., Liu, R., Zhao, J. Z., Shung, K. K., Nguyen, H., Chia, R., Dentinger, A., Wildes, D., Thomenius, K. E., Mahajan, A., Shivkumar, K., Kim, K., O’Donnell, M., & Sahn, D. J. (2008). The Acoustic Lens Design and in vivo Use of a Multifunctional Catheter Combining Intracardiac Ultrasound Imaging and Electrophysiology Sensing. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 55(3), 602-618. Stephens, D. N., O’Donnell, M., Thomenius, K., Dentinger, A., Wildes, D., Chen, P., Shung, K. K., Cannata, J., Khuri-Yakub, P., Oralkan, O., Mahajan, A., Shivkumar, K., & Sahn, D. J. (2009). Experimental Studies with a 9F Forward-Looking Intracardiac Imaging and Ablation Catheter. AIUM J. Ultrasound Med., 28(2), 207-215. Stevenson, G., Wilson, J., Wilkinson, J., Norman, G., & Goodacre, R. L. (1992). Pain following colonoscopy: Elimination with carbon dioxide. Gastrointest. Endosc., 38(5), 564-567. Szabo, T. L. (2013). Diagnostic Ultrasound Imaging: Inside Out. 2 nd Edition. Oxford, UK. Elsevier Academic Press. Turnbull, D. H., Starkoski, B. G., Harasiewics, K. A., Semple, J. L., From, L., Gupta, A. K., Sauder, D. N., & Foster, F. S. (1995). 40-100 MHz B-scan ultrasound backscatter microscope for skin imaging. Ultrasound Med. Biol., 21(1), 79-88. Verasonics. (2015). The Vantage Research Ultrasound Systems. Retrieved from: http://verasonics.com/index.php/vantage-systems/ Verasonics. (2015). Verasonics Custom Transducer Connector Options. Retrieved from: http://verasonics.com/index.php/backshell-connector-kit/ Verasonics. (2017). Vantage Sequence Programming Manual, Verasonics, Inc., Kirkland, WA, USA, 84-87, 167-169. WebMD. (2015). Transrectal Ultrasound. Retrieved from: http://www.webmd.com/digestive- disorders/transrectal-ultrasound World Health Organization. (2015). Cardiovascular Diseases (CVDs), Fact sheet no. 317, 2015. Retrieved from: http://www.who.int/mediacentre/factsheets/fs317/en/ Yamada, T., Alpers, D. H., Kalloo, A. N., Kaplowitz, N., Owyang, C., & Powell, D. W. (2011). Principles of Clinical Gastroenterology. Chichester, West Sussex. Wiley-Blackwell. 187 Yang, Y., Li, X., Wang, T., Kumavor, P. D., Aguirre, A., Shung, K. K., Zhou, Q., Sanders, M., Brewer, M., & Zhu, Q. (2011). Integrated optical coherence tomography, ultrasound and photoacoustic imaging for ovarian tissue characterization. Biomedical Optics Express, 2(9), 2551-2561. Yakish, S. J., Narula, A., Foley, R., Kohut, A., & Kutalek, S. (2015). Superior Vena Cava Echocardiography as a Screening Tool to Predict Cardiovascular Implantable Electronic Device Lead Fibrosis. J. Cardiovasc. Ultrasound, 23(1), 27-31. Yoon, S., Kim, M. G., Williams, J. A., Yoon, C, Kang, B. J., Cabrera-Munoz, N. E., Shung, K. K., & Kim, H. H. (2015). Dual-element needle transducer for intravascular ultrasound imaging. J. Med. Imag., 2(2), 027001-1 – 027001-9. Yoon, S., Williams, J. A., Kang, B. J., Yoon, C., Cabrera-Munoz, N. E., Jeong, J. S., Lee, S. G., Shung, K. K., & Kim, H. H. (2015). Angled-focused 45MHz PMN-PT single element transducer for intravascular ultrasound imaging. Sensors and Actuators A: Physical. 228, 16-22. Zacharakis, E., Freilich, S., Rekhraj, S., Athanasiou, T., Paraskeva, P., Ziprin, P., & Darzi, A. (2007). Transanal endoscopic microsurgery for rectal tumors: The St. Mary’s experience. American J. Surg., 194(5), 694-698. Zagzebski, J. A. (1996). Essentials of Ultrasound Physics. St. Louis, MO. Mosby Elsevier Science. Zagzebski, J. A., Banjavic, R. A., Madsen, E. L., & Schwabe, M. (1982). Focused transducer beams in tissue-mimicking material. J. Clin. Ultrasound., 10(4), 159-166.
Abstract (if available)
Abstract
This dissertation describes the design, fabrication, and testing of two different high-frequency ultrasound array probes for medical imaging applications. State-of-the-art piezoelectric materials, fine spatial scale high-frequency composites, modeling tools, custom-engineered electrical and mechanical components, as well as novel fabrication techniques were employed to maximize the performance of the arrays. ❧ The first probe consists in a 15-MHz, side-looking, miniature linear phased array for colorectal robotic surgery. The array is intended to be integrated into a robotic surgery probe for the real-time intraoperative fusion of Endorectal Ultrasound (ERUS) and endomicroscopy to assist in the assessment of colorectal cancer surgical margins during Transanal Endoscopic Microsurgery (TEM) procedures. The measured performance and imaging evaluation confirmed that the fabricated array provides enough penetration (≥ 8.1 mm), peak-to-peak sensitivity (≥ 100 mV), and spatial resolution (90 μm axially / 420 μm laterally) for the overall structure detection of colorectal cancerous tissue. ❧ The second probe consists in a 30-MHz, forward-looking, miniature linear phased array for Intravascular Ultrasound (IVUS) imaging and intraluminal vessel navigation during Peripheral Arterial Disease (PAD) angioplasty procedures. The measured performance and imaging evaluation confirmed that the fabricated array provides enough depth of view (≥ 5.0 mm), peak-to-peak sensitivity (≥ 25 mV), and spatial resolution (65 μm axially / 215 μm laterally) for the overall detection of structures lying ahead of the catheter.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Design and development of ultrasonic array transducers for specialized applications
PDF
Highly integrated 2D ultrasonic arrays and electronics for modular large apertures
Asset Metadata
Creator
Cabrera-Munoz, Nestor Emmanuel
(author)
Core Title
Miniature phased-array transducer for colorectal tissue characterization during TEM robotic surgery; and, Forward-looking phased-array transducer for intravascular imaging
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Biomedical Engineering
Publication Date
08/01/2018
Defense Date
12/11/2017
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
connectors,electrical interconnects,finite-element analysis,flexible printed circuits,high-frequency imaging,medical imaging,medical transducers,micro-coaxial cabling,miniature transducers,OAI-PMH Harvest,piezoelectric materials,piezoelectric transducers,transducer modeling,transmission lines,ultrasonic imaging,ultrasonic transducer arrays
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Shung, Koping Kirk (
committee chair
), Chen, Mike Shuo-Wei (
committee member
), Zavaleta, Cristina (
committee member
)
Creator Email
cabreram@usc.edu,nestor.ecm@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-467071
Unique identifier
UC11266141
Identifier
etd-CabreraMun-5990.pdf (filename),usctheses-c40-467071 (legacy record id)
Legacy Identifier
etd-CabreraMun-5990.pdf
Dmrecord
467071
Document Type
Dissertation
Rights
Cabrera-Munoz, Nestor Emmanuel
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
connectors
electrical interconnects
finite-element analysis
flexible printed circuits
high-frequency imaging
medical imaging
medical transducers
micro-coaxial cabling
miniature transducers
piezoelectric materials
piezoelectric transducers
transducer modeling
transmission lines
ultrasonic imaging
ultrasonic transducer arrays