University of Southern California Dissertations and Theses

Ultrasound neuromodulation and its applications for noninvasive vision restoration

(USC Thesis Other)

Ultrasound neuromodulation and its applications for noninvasive vision restoration

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Ultrasound Neuromodulation and its Applications for Noninvasive Vision Restoration

By

Gengxi (Alex) Lu

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)

December 2022

Copyright 2022 Gengxi Lu

ii

Dedication

To my beloved
Parents
Guiqi Lu & Yuemei Gong

Fiancée
Wanyi Dong

Copyright 2022 © Gengxi Lu

iii

Acknowledgments
This thesis would not have been possible without the help of so many people in
different ways. When looking back at this five-year journey, I feel so lucky and
happy to meet and be in contact with lots of wonderful people who have contributed
to my personal growth and professional development in medical ultrasound.
I am sincerely grateful to my advisor Professor Qifa Zhou, for his guidance,
mentoring, and support during my Ph.D. study at NIH Ultrasonic Transducer
Resources. His patience, knowledge, and trust encouraged and supported me all
these five years. What I learned from Dr. Zhou is not only knowledge but also a life
philosophy.
I would also like to express my sincere gratitude to my dissertation committees:
Dr. Qifa Zhou, Dr. Michael Khoo, and Dr. Mark S. Humayun, for their insightful
suggestion during my preparation of this dissertation.
I would like to greatly acknowledge my collaborators, Dr. Mark S. Humayun, Dr.
Biju Thomas, Dr. Zhongping Chen, and Dr. Ronald H. Silverman, for sharing their
expertise and clinical perspectives for this research.
I am heartily thankful to all my colleagues and friends that I met at USC, Xuejun
Qian, Laiming Jiang, Zeyu Chen, Ruimin Chen, Robert Wodnicki, Mingyue Yu,
Haocheng Kang, Runze Li, Yizhe Sun, Yushun Zeng, Chen Gong, Haotian Lu,
Junhang Zhang, Adnan Rayes, and Wenxuan Jiang. We have been working together
for many years, and my research life became enjoyable and memorable with the
support and company of all of them.
Last but not least, I would like to thank my parents, who raised me and gave me
unconditional love, as well as my fiancée Wanyi Dong, who has always been
supportive and stood by my side through all the good times and bad.

iv

Table of Contents

Dedication ................................................................................................................................................... ii
Acknowledgments ..................................................................................................................................... iii
List of Tables ............................................................................................................................................ vii
List of Figures .......................................................................................................................................... viii
Abstract ......................................................................................................................................................xiii
Chapter 1 Introduction ............................................................................................................................... 1
1.1 Neuromodulation and Vision Restoration ........................................................................ 1
1.1.1 Applications .................................................................................................................... 1
1.1.2 Retinal Prostheses ........................................................................................................... 1
1.1.3 ON and VC Prostheses ................................................................................................... 3
1.1.3 Emerging Noninvasive Neuromodulation Technologies ............................................... 4
1.2 US Effects ............................................................................................................................. 6
1.3 Motivations and Objectives ................................................................................................... 8
1.4 Outline ................................................................................................................................... 8
Chapter 2 Non-invasive Ultrasound Retinal Stimulation for Vision Restoration............................ 11
2.1 Introduction ......................................................................................................................... 11
2.2 Methods ............................................................................................................................... 13
2.2.1 Experimental Setup....................................................................................................... 13
2.2.2 Animal Preparation ....................................................................................................... 14
2.2.3 Ultrasound Transducer ................................................................................................. 15
2.2.4 Ex-vivo Study ................................................................................................................ 15
2.2.5 Stimulation Parameters and Patterns ............................................................................ 15
2.2.6 Data Processing ............................................................................................................ 16
2.2.7 Histology Analysis ....................................................................................................... 16
2.2.8 Finite Element Analysis Simulation ............................................................................. 17
2.2.9 Statistical Analysis ....................................................................................................... 17
2.3 Results ................................................................................................................................. 18

v

2.3.1 Ultrasound Transducer and Power Calibration ............................................................ 18
2.3.2 US-evoked Neuron Response in Rat in vivo ................................................................ 19
2.3.3 US Parameters on Response Pattern ............................................................................. 23
2.3.4 Spatiotemporal Resolution and Frame Rate of US stimulation .................................... 24
2.4 Discussion and Conclusion ................................................................................................. 27
Chapter 3 2D-array Based Ultrasonic Retinal Prosthesis and Its Frequency Dependent
Efficiency .................................................................................................................................................. 33
3.1 Introduction ......................................................................................................................... 33
3.2 Materials and Methods ........................................................................................................ 33
3.2.1 Ultrasound 2D Array Design, Fabrication .................................................................... 33
3.2.2 Neuron-recording Experiment Setup ............................................................................ 37
3.2.3 Water-licking Behavioral Experiment Setup ................................................................ 39
3.2.4 Algorithm for Pattern Generation ................................................................................. 40
3.2.5 Animal Preparation ....................................................................................................... 40
3.2.6 Ultrasound Imaging Guide for Retinal Stimulation ..................................................... 41
3.2.7 Single-element Transducers Design and Fabrication ................................................... 42
3.3 Results ................................................................................................................................. 43
3.3.1 Ultrasound Imaging-guided Retinal Stimulation .......................................................... 43
3.3.2 Ultrasound Pattern Generation and Stimulation ........................................................... 44
3.3.3 Frequency-dependent Stimulation Efficiency .............................................................. 46
3.3.4 Safety Investigation ...................................................................................................... 48
3.3.5 Water-licking Behavioral Test ...................................................................................... 50
3.4 Discussion and Conclusion ................................................................................................. 52
Chapter 4 Transcranial Focused Ultrasound for Non-invasive Neuromodulation of the Visual
Cortex ......................................................................................................................................................... 55
4.1 Introduction ......................................................................................................................... 55
4.2 Materials and Methods ........................................................................................................ 57
4.2.1 Ultrasound Transducer and Waveform ......................................................................... 57
4.2.2 Animal Preparation ....................................................................................................... 58
4.2.3 Recording of VC-Evoked Potential Activities ............................................................. 58
4.2.4 Experimental Design .................................................................................................... 59
4.3 Results ................................................................................................................................. 59
4.3.1 Ultrasound Field ........................................................................................................... 59

vi

4.3.2 VC Responses ............................................................................................................... 61
4.4 Discussion and Conclusion ................................................................................................. 63
Chapter 5 Summary and Future Work ................................................................................................... 68
5.1 Summary ............................................................................................................................. 68
5.2 Future Work ........................................................................................................................ 68
5.2.1 Wearable High-frequency Ultrasound Array-based Retinal Prosthesis ....................... 69
5.2.2 High-frequency US VC Stimulation............................................................................. 69
5.2.3 Study of Biophysical Mechanism ................................................................................. 69
BIBLIOGRAPHY .................................................................................................................................... 70

vii

List of Tables
Table 1-1. Summary of Non-invasive Neuromodulation Modalities…………………………...7
Table 2-1. List of acoustic and thermal parameters of water and organs of eye……………..….17
Table 2-2. The relationship between the input voltage (before 50 dB gain) and ultrasound
pressure, intensity, and mechanical index (MI)……………………………………………….…18
Table 3-1. Summary of Ultrasound Retinal Stimulation Results ………………………………...36

viii

List of Figures
Figure 1-1. Schematic diagram of the Argus II retinal prosthesis system. (Source figure is
from Weiland et al. 2005 review paper) ……………………........................................................ 2
Figure 1-2. Diagrammatic sketch of a cortical visual prosthetic system. (Source figure is
from Beauchamp et al. Cell 181.4 (2020): 774-783) .….……....................................................... 3
Figure 1-3. The illustration of acoustic cavitation effects and acoustic radiation force. .............. 6
Figure 2-1. Schematic diagram of the experiment setup. (a) The retina stimulation part.
A focused single-element transducer was used to generate ultrasound waves targeting the
retina. Retinal neurons were excited and generated neural signals transmitting through the
optic nerve to the brain. (b) The brain recording part. A multielectrode array (MEA) was
inserted into the contralateral superior colliculus (SC) or visual cortex (VC)…………......... 12
Figure 2-2. The simulated acoustic field: (a) X-Z plane and (b) X-Y plane. (c) Measured
amplitude of pressure and MI. ……………................................................................................. 19
Figure 2-3. The simulated acoustic field: (a)&(b) X-Z plane and (d) X-Y plane. (c)
Simulated temperature increase. (e) The axial acoustic pressure distribution along the
z-axis. (f) The curve of temperature changes over time. ............................................................. 20
Figure 2-4. Representative US-evoked neuron responses recorded from SC. (a) Light
response from normal sighted rat. (b) US response from normal sighted rat. (c) Light
response from blind rat. (d) US response from blind rat. The first row shows filtered
signals by bandpass filter between 500 Hz and 7000 Hz. The second row shows the average
spike counts per 5 ms of all channels. In the third row, a representative channel was selected
to demonstrate the spike counts curve. The deep blue shows the average value, and the baby
blue shows the standard deviation. ………………….............................................……………..21
Figure 2-5. Representative US-evoked neuron responses recorded from VC. (a) Light
response from normal sighted rat. (b) US response from normal sighted rat. (c) Light
response from blind rat. (d) US response from blind rat. The first row shows filtered signals
by bandpass filter between 500 Hz and 7000 Hz. The second row shows the average spike
counts per 5 ms of all channels. In the third row, a representative channel was selected to

ix

demonstrate the spike counts curve. The deep blue shows the average values, and the baby
blue shows the standard deviation. .............................................................................................. 22
Figure 2-6. US-evoked neural responses change along with the US amplitude (a) and
duration (b). The statistical analysis of the effect of US parameters. (c) Response intensity
changes along US duration. (d) Response duration changes along the US duration. (e)
Response intensity changes along US amplitude. (f) Response duration changes along US
amplitude. (g) The comparison of responses between normal rats and retina-degenerated
rats. …………………................................................................................................................... 23
Figure 2-7. The effect of pulsed ultrasound stimulation. (a) A schematic diagram of pulsed
US waveform. (b) Representative responses with different duty cycles. (c) Statistical
analysis of the effect of duty cycle. ............................................................................................. 25
Figure 2-8. (a) Schematic diagram of the surface of SC and the MEA. (b) A representative
response mapping to show the resolution of the response. (c) Different positions of SC were
activated by stimulating different areas. (d) Repetitive responses from stimulations with a
PRF of 5 Hz. (e) Latency difference between light response and US responses from normal
sighted rats and retina degenerated rats. ………...………………………………………………26
Figure 2-9. Representative histology results of US stimulated retinas from both a
normal-sighted rat and a RCS blind rat. …................................................................................... 29
Figure 3-1. (a) The layered structure of an element in the array. (b) The photo of the
fabricated array together with the interposer and PCB board………........................................... 34
Figure 3-2. Schematic diagram of ultrasound retinal stimulation system. Top: The workflow
of 2D-array ultrasound retina stimulation with dynamic patterns. Calculated amplitude and
phase distribution are applied to the 2D array. Then the generated patterns are used to
stimulate the retina in vivo. MEA was used to map the neuron responses from the brain. Bot:
Verasonics ultrasound system. The detailed structure of one element in the array. The
ultrasound evoked neuron responses with a “C” pattern. ……....………………...……............. 37
Figure 3-3. The design of the new MEA to cover the whole surface of the SC………….......... 38

x

Figure 3-4. The schematic diagram of the water-licking behavioral experiment. The Arduino
is used to control and synchronize the LED flashlight, water pump, and ultrasound
stimulation. A night vision camera is used to monitor the water-licking behavior of rats. ......... 39
Figure 3-5. Top row: the hydrophone measured the frequency responses of all three
transducers. Bot row: the 2D mapping of the focus of each transducer at the focal plane.
Higher-frequency transducers have finer focal points. The FWHM is 640 µm of the 3.1-MHz
transducer, 510 µm of the 5.4-MHz transducer, and 110 µm of the 20-MHz transducer. ……...42
Figure 3-6. 4.5-MHz ultrasound imaging using the 2D array. Although the image quality
is relatively poor, eyeball structures still can be distinguished. ................................................... 43
Figure 3-7. (a) Amplitude distribution of 2D array for “C” pattern. (b) Phase distribution of
2D array for “C” pattern. (c) Simulated generated “C” at the pattern plane (20 mm away from
the array surface). (d) Mapped neuron responses at SC of “C” pattern. (e) Mapped neuron
responses at SC of “V” pattern. (f) Mapped neuron responses at SC of “S” pattern. .................. 44
Figure 3-8. (a) The frequency spectrum measured by hydrophone of the single-element
transducers used for mechanism study. Transducers were driven at different frequencies, and
their output pressures were calibrated with the spectrum. (b) Frequency-related pressure
threshold. (c) Frequency-related mechanical index threshold. .................................................... 45
Figure 3-9. Fundus imaging and OCT of the retina (a) before and (b) after ultrasound
stimulation..................................................................................................................................... 47
Figure 3-10. ERG results before and after ultrasound stimulation. (a)&(b) show the ERG
signals after two weeks of ultrasound stimulation using a normal intensity. (c)&(d) show
the ERG of the untreated eye (control group) before and after the stimulation. (e)&(f) are
the negative controls, showing the ERG change of the eye stimulated by excessive
ultrasound intensity (two times stronger than the normal intensity). ........................................... 48
Figure 3-11. Representative histology results of stimulated and control eyeball. (a) H&E
staining, (b) GFAP staining, (c) CD68 staining. .......................................................................... 49
Figure 3-12. Water-licking behavior tests show that rats have similar behavioral responses to
light and ultrasound retinal stimulations. (a) Licking rates change along the time in each trial:

xi

before, during, and after the stimulation. (b) The anticipatory licking rate increased along the
trial number. (c) Raster plot of licking behavior during each trial. (d) The summary and
statistical analysis of licking rates in three stages……………...……………………………...... 51
Figure 3-13. Water-licking behavioral tests of blind rats. (a) Anticipatory licking rates in the
eight days of training. The licking rates increase with more learning. The black line shows the
learning curve of normal rats, and the blue line shows the learning curve of normal rats. (b)
Licking action happens before, during, and after the stimulation. Blind and normal rats show
similar curves, indicating the same learning behavior.................................................................. 52
Figure 4-1. (a) Schematic diagram of the experimental system. (b) The surface topography
of the cortical areas of the left hemisphere of the rat. Each background grid has a length of 2
mm. The primary visual cortex, area 17, is shown stippled. The exact position of the electrode
is shown by the red point. The ultrasound focal area is shown by the orange circle. (c) Top:
Simulation results of the spatial distribution of acoustic intensity. Bottom: Simulation results
of the ultrasound-induced temperature increasing and its spatial distribution. ........................... 57
Figure 4-2. Representative results evoked by continuous ultrasound stimulation from one
normal rat (a-d) and one blind rat (e-h). Gray lines show eight-times records, and red lines
are the averaged records. Light stimulation was used to test the rats’ visual responsiveness.
Normal rats responded to light (a), while blind rats failed to show any light responses (e).
VC potential baseline recorded during control experiments (when the transducer was focused
in a different direction) showed no responses in both rat groups (b)&(f). (c)&(g) show the
responses with 15ms ultrasound stimulation (shown by the red line) from group 1. (d)&(h)
show the responses from group 2.........................................................................................…..... 60
Figure 4-3. Comparison of response amplitude in groups 1&2 showing blind rats have
significantly stronger VC responses to US stimulation (p<0.001, two-sample t-test). In
experimental group 1, normal rats showed a response amplitude of 1.88±0.04 µV, whereas
blind rats had 4.87±0.38 µV. In group 2, the response amplitudes were 1.69±0.18 µV for
normal rats and 3.84±0.14 µV for blind RCS rats. ...................................................................... 61
Figure 4-4. VC low-frequency responses to pulsed US stimulation with high PRF. US
stimulation duration was 30 ms, which is shown by the red bar at the bottom. All signals

xii

were averaged 512 times. Signals were recorded with different stimulation parameters: (a)
1-ms pulse in every 2 ms (500 Hz PRF and 50% DC); (b) 0.5-ms pulse in every 2 ms (500
Hz PRF and 25% DC); (c) 1-ms pulse in every 3 ms (333.3 Hz PRF and 33.3% DC); (d)
2-ms pulse in every 5 ms (200 Hz PRF and 40% DC); (e) 5-ms pulse in every 10 ms (100
Hz PRF and 50% DC). Most responses (a)-(d) had a similar waveform which was labeled by
P1, P2, N1, and N2 in (a). …………………………………………………………………….... 62

xiii

Abstract
Neuromodulation is a technology that directly stimulates neurons and modulates neural
activities through the targeted delivery of a stimulus to specific neurological sites in the body.
Since the first clinical success of the use of electrical stimulation for pain relief, neuromodulation
has been widely employed in prostheses or neuropathic disease treatment.
Electrical and chemical stimulations are the most conventional techniques for
neuromodulation. They are widely used in clinics and research, but they also have limitations,
such as invasiveness and resolution. In recent decades, different types of new neuromodulation
technologies have been investigated, including transcranial magnetic stimulation (TMS),
transcranial direct current stimulation (tDCS), optogenetics, sonogenetics, and ultrasound (US)
stimulation. Of these emerging methods, US stimulation has unique advantages, including
noninvasiveness, deep penetration, high resolution, and cost-effectiveness.
The work presented in this dissertation investigates the feasibility of US stimulation for visual
prostheses. It was demonstrated that focused US stimulation could evoke neural activities in the
retina and visual cortex. Experiments were conducted in vivo on normal-sighted Long Evans rats
and retinal degenerative Royal College of Surgeons rats. Parameters of US such as amplitude and
duration were optimized for stimulation by quantifying the evoked neural responses. The
spatiotemporal resolution of US stimulation was investigated by mapping the neuron response with
a multi-electrode array (MEA) and changing US waveforms. A 2D array-based US stimulation
with arbitrary patterns was also presented. Water-licking tests were conducted to demonstrate that
animals’ behavioral responses to US retinal stimulation are similar to those of light stimulation. In
addition, the safety and physical mechanisms of US stimulations were investigated by measuring
the amplitude of US pressure and the US-induced heating effect. Eye imaging and histology were
also carried out to ensure there was no morphological damage to the stimulated area. The results
indicated the capability of the US stimulation to evoke neuronal activity. US-based visual
prosthesis could be a promising technology for blindness treatment. Furthermore, the favorable
prospect of the ultrasonic retinal prosthesis in the next step is predicted. Our future study will also
examine the biological mechanism of US stimulation via in vitro and single-cell stimulation.

1

Chapter 1 Introduction
1.1 Neuromodulation and Vision Restoration
1.1.1 Applications
Neuromodulation is a technology that directly stimulates neurons and modulates neural
activities through the targeted delivery of a stimulus to specific neurological sites in the
body. Electrical stimulation is the most common technique clinically used for treating
neuropathic diseases. Since Shealy et al. implanted the first neurostimulator in 1967 [1],
electrical spinal cord stimulation has been used to treat neuropathic pain for over 50 years
[2]. Several other neuropathic diseases are related to the brain. Deep brain stimulation
(DBS) is an FDA-approved electrical neuromodulation method [3]. DBS can be used for
treating Parkinson’s disease [4, 5], epilepsy [6, 7], depression [8], obsessive–compulsive
disorder [9, 10], morbid obesity [11], Alzheimer’s disease [12], and dementia [13] by
inserting electrodes into different subcortical areas and using various stimulation
waveforms
Prosthesis is another essential application field of electrical stimulation and includes
auditory prosthesis [14-17], visual prosthesis [18-22], and bladder control [23, 24].
Prostheses can be implanted to stimulate various areas along the related neural pathway.
For example, auditory prostheses have three major types: cochlear implants [14], auditory
brain stem implants [14, 15], and auditory midbrain implants [16, 17].
1.1.2 Retinal Prostheses
Similar to auditory prostheses, visual prostheses have been implanted on the retina [18,
19, 22], optic nerve (ON) [21], and visual cortex (VC) [20]. The targeted region of
prostheses can be chosen based on the patients’ condition. Stimulating the low-level neural
system (retina) can benefit from the natural information processing along the visual
pathway. However, stimulating the high-level neural system is necessary for patients with
a damaged visual pathway, such as the ON damaged by glaucoma or accident trauma.
As presented in Fig. 1-1 [22], retinal prosthesis functions as an integral system that
comprises an image acquisition device, image processor, stimulator chip, and electrode
array to electrically stimulate neurons. An image acquisition device, such as a video camera,

2

captures images from the visual field and passes them to a specialized computer to translate
these images to stimulation patterns for the electrode array. The implanted stimulator
generates the stimulus pattern, which is then delivered to the multi-electrode array (MEA)
positioned near the retina. The power and data transmission between the external parts of
the device and the implant takes the form of radio frequency (RF) and/or an optical link
that is minimally invasive. Three major types of retinal implants have been developed:
epiretinal prostheses, anchored to the inner surface of the retina; subretinal prostheses,
embedded between the retina and the RPE/choroid; and suprachoroidal prostheses,
implanted between the choroid and the sclera. Using electrical stimulation to directly elicit
neural activity at the inner retina, retinal prostheses promise the best near-term strategy to
provide partial vision restoration in patients with retinal degenerate (RD) [22, 25]. The
above retinal stimulation strategy has resulted in the clinical implementation of many
retinal implants, including Argus II epiretinal prosthesis (Second Sight Medical Products,
Sylmar, CA, USA) [26], Alpha IMS/AMS photodiode-based subretinal prosthesis (Retinal
Implant AG, Reutlingen, Germany) [27], and PRIMA photovoltaic-driven subretinal

Figure 1-1. Schematic diagram of the Argus II retinal prosthesis system. (Source figure is
from Weiland et al. 2005 review paper)

3

prosthesis (Pixium Vision, Paris, France) [28]. However, implanted devices require surgery
and pose other possible risks due to invasiveness.
1.1.3 ON and VC Prostheses
Unlike retinal prosthesis, which has been FDA-approved, research on ON and VC
prostheses is still ongoing. Several studies have investigated the potential of stimulating
ON to restore vision with spiral cuff electrodes [29-31]. However, to the author’s best
knowledge, there is no commercial device for ON stimulation. One major disadvantage of
ON prosthesis is the compact coaxial structure of ON, which makes it extremely difficult
to reach meaningful visual patterns and good spatial resolution.
The currently available approach for VC prosthesis described above is based on
electrical stimulation of the cortical neurons. The cortical neurons respond to the
electrical stimulation generated by the MEAs implanted inside the brain through an
invasive surgical procedure. Recently, Second Sights, Inc., developed a visual cortical
prosthesis (Fig. 1-2). This so-called Orion system was designed to convert images

Figure 1-2. Diagrammatic sketch of a cortical visual prosthetic system. (Source figure is
from Beauchamp et al. Cell 181.4 (2020): 774-783)

4

captured by a miniature video camera mounted on a pair of glasses into a series of
electrical pulses wirelessly transmitted to an array of electrodes placed on the surface of
the VC [20]. Although the use of a cortical visual prosthesis has the advantage of
avoiding device implantation in the delicate retinal tissue, most issues associated with
eye-implanted devices could persist when MEAs are implanted in the brain. Significant
issues of implanted cortical devices include biocompatibility, encapsulation, power
requirements, electrode degradation, and interference with residual vision [25].
Complications may also occur due to surgical invasiveness, including risk of seizure,
chronic inflammation, and neurodegeneration [32]. MEAs can provide only limited
spatial resolution due to the small number of stimulating electrodes. To improve
stimulation efficacy, a larger number of electrodes can be used to mediate spatial
stimulation patterns capable of generating complex spatiotemporal percepts required for
a useful visual perception [33, 34]. The number of electrodes needed to be implanted in
the VC to achieve useful visual sense in a profoundly blind subject is estimated to be
approximately 625 [35]. However, there is also the inherent difficulty of complex and
invasive implantation techniques related to many electrodes needed for high resolution.
1.1.3 Emerging Noninvasive Neuromodulation Technologies
To address the invasive nature of electrode-based neurostimulator, different types of
new noninvasive neuromodulation technologies have been recently proposed and
assessed, including interfering electric fields [36], repetitive transcranial magnetic
stimulation (TMS) [37], transcranial direct current stimulation (tDCS) [38],
optogenetics [39, 40], sonogenetics [41, 42], and US stimulation [43-46].
TMS and tDCS are noninvasive techniques that use magnetic field and electric
current to modulate the neuronal activities in the target area. However, they suffer from
a relatively low resolution on the order of a few centimeters. It is also challenging to use
them for targeting deep subcortical areas.
Optogenetics uses light to control the neurons genetically modified to express light-
sensitive ion channels. A recent work from Nelidova et al. successfully enabled retinal
degenerative mice to perform a learned near-infrared light-driven behavior. By
controlling the gene-engineered ion channels, researchers can tune the response to

5

various wavelengths and the response thresholds [47]. The benefits of optogenetics
include noninvasiveness after gene expression, high resolution, and cell selectivity.
However, in the case of practical applications, it suffers from poor penetration and
complicated gene engineering.
Sonogenetics shares similar principles and benefits to optogenetics, except that the
expressed ion channels are sensitive to US stimulation. Mechanosensitive ion channels
[41, 48] and temperature-sensitive ion channels [49] are commonly used in sonogenetics.
Better than the light, transcranial ultrasound can be delivered to the deep brain
noninvasively. Still, complicated gene engineering is needed in sonogenetics, and the
initial gene expression is an invasive method.
Of these emerging methods, US stimulation has unique advantages. US
neuromodulation was pioneered by Edmund Harvey in 1929 in ex vivo frogs [50]. It was
demonstrated that high-intensity US stimulation could evoke muscle contraction. About
30 years later, the first in vivo US neuromodulation result based on the visual pathways
was presented by F. J. Fry [51]. This study reported that US stimulation of the cat’s
lateral geniculate nuclei for 20–120 s significantly inhibited the light-evoked potentials.
This suppression was completely recovered 30 min after the US exposure. Table 1-1
presents the comparison of US stimulation with other stimulation modalities. The µm-
level resolution and cm-level penetration depth make US a promising candidate for next-
generation noninvasive visual prosthesis. Unlike sonogenetics, US stimulation uses the
neurons’ natural ion channels, which makes it easy to be used in clinical applications.
Furthermore, as a noninvasive technique, US stimulation has great resolution (as fine as
tens of micrometers) and penetration (centimeters to cover the whole brain). The current
disadvantages of US stimulation are its unclear mechanism and neuronal dependence.
Some studies have investigated the mechanism of US stimulation on which ion channels
play the most critical roles. However, no conclusions can be drawn for now. Furthermore,
because no gene expression is used, it is impossible to stimulate the neurons that lack
US-related ion channels. The following sections will discuss more details about the US
stimulation mechanism.

6

1.2 US Effects
US has three major effects that could be the mechanism of US stimulation: cavitation,
acoustic radiation force (ARF), and thermal effects. The cavitation effect is the interaction
between US waves and bubbles, where bubbles can be generated if the acoustic pressure
has sufficiently negative peaks. The size of the bubble oscillates as the localized pressure
changes sinusoidally. Transient cavitation occurs when the size expansion is at least double,
at which point the bubble violently collapses, causing a destructive event [52]. In stable
cavitation, the size change is smaller, and the bubbles do not burst, which is hypothesized
to produce stable neuromodulation [53].
The phenomenon of ARF generation results from the propagation of acoustic waves
through a dissipative medium. It is induced by a transfer of momentum from the wave to
the medium, arising either from absorption or scattering. In general, the contribution to the
ARF by soft tissue can be neglected. Thus, in an absorbing medium and under plane wave
assumptions, the ARF can be defined as follows:
=
2

where denotes the ARF; , the absorption coefficient of the medium; , the speed of
sound in the medium; and , the temporal-average intensity at a given point in space [54,
55]. For a focused acoustic beam, a radiation force is applied throughout the focal region
of the acoustic beam. The region of excitation (ROE) defines the focal zone of the ARF.
The thermal effect of the US is caused by the absorption of ultrasonic vibration, in which
the mechanical energy is converted into heat. The US-induced thermal effect is an
Figure 1-3. The illustration of acoustic cavitation effects and acoustic radiation force.

7

Table 1-1. Summary of Non-invasive Neuromodulation Modalities
Modality Interfering electric fields transcranial
magnetic
stimulation
transcranial
direct current
stimulation
Optogenetics Sonogenetics Ultrasound
Principle deliver multiple electric fields
at different frequencies. The
modulated electric field
envelop are used to stimulate
neurons in the interference
region.
induce electrical
currents to stimulate
neurons using
pulsating magnetic
fields that are
generated outside
the body.
deliver low-
amplitude (usually
no more than 2
mA) electric
current between
electrodes (anode
and cathode).
gene expression +
light stimulation
gene expression
+ ultrasound
stimulation
(ultrasonic
mechanical or
thermal effects)
ultrasound
stimulation
(ultrasonic
mechanical or
thermal effects)
Resolution several mm to cm
(Medium)
several cm
(Low)
several cm to tens
of cm
(Low)
hundreds of nm to
um
(Highest, Single
neuron resolution)
µm to mm
(High)
µm to mm
(High)
Depth several mm to centimeters
(Medium)
several cm
(High)
several cm to tens
of cm
(High)
several mm
(Low)
several cm
(High)
several cm
(High)
Cell-type
selectivity
No No No Yes Yes No

8

accumulative phenomenon determined through the acoustic intensity, absorption
coefficient, duration, and duty cycle.
1.3 Motivations and Objectives
In general, the proposed study has two objectives: to develop noninvasive
therapeutics based on US stimulation, especially for novel visual prostheses, and to
examine the mechanism of US stimulation.
At present, most clinical neuromodulations are still limited to invasive electrical
stimulation, which requires the implantation of electrodes. For visual prostheses, the
difficulty of the invasive implantation of electronic devices, limited number of
electrodes, high surgical costs, and potential side effects of surgery remain unsolved. A
noninvasive visual prosthesis is an unmet requirement for blind patients. US stimulation
as a promising noninvasive neuromodulation technology has been investigated and
demonstrated by many studies. Some studies have indicated that US can evoke localized
neural activities by stimulating the retina [53] and VC [43]. However, in vivo
demonstrations of vision restoration in blind animals are still lacking. In addition, most
in vivo studies only show the capability of US stimulation to evoke neural activities. The
spatiotemporal resolution of US stimulation has never been investigated, which is
significant for practical applications, especially for visual prosthesis.
Investigation of the mechanism of US stimulation could have significant meanings in
basic neuroscience and clinical applications. Different mechanosensitive and
temperature-sensitive ion channels have been found in the cells. But their roles in US
stimulation are still unclear. Understanding the mechanism could also help optimize the
US parameters to modulate neurons more efficiently.

1.4 Outline
This thesis is structured as follows:
Chapter 1 introduces the concept and development of neuromodulation and its
applications in medical treatments. The development of visual prostheses is also
specifically introduced. Emerging neuromodulation technologies that have the potential to
overcome the limitations of conventional electrical stimulation are also discussed. Then,

9

noninvasive US stimulation is presented, and the physical mechanisms of US effects are
listed. Finally, the motivation and possible clinical requirements for developing US-based
noninvasive visual prostheses are demonstrated. The scientific and practical significance
of examining the mechanism of US stimulation is also addressed.
Chapter 2 suggests the use of US stimulation in the retina for vision restoration. It is
demonstrated that US stimulation on either normal-sighted or retinal degenerative blind
rats in vivo can evoke neuronal activities in the visual pathway to the brain, demonstrating
ultrasound as an effective neuromodulation approach. The implemented 3-MHz ultrasound
transducer exhibited both good spatial resolution of 300 µm and temporal resolution of 5
Hz on spike activities of the superior colliculus. Various stimulation parameters are tested,
and US-evoked neuron responses are evaluated in detail. It was further demonstrated that
ultrasound neuromodulation on the retina in vivo is a safe manner. All the results indicate
that US retinal stimulation is a promising technology as a novel and noninvasive visual
prosthesis for translational studies on patients with blindness.
Chapter 3 proposes and validates a 2D array-based ultrasonic noninvasive retinal
prosthesis. A 16 × 16 ultrasound 2D array with a center frequency of 4.5 MHz was
designed and fabricated to dynamically stimulate the retina. Dynamic acoustic patterns are
generated by controlling the amplitude and phase of each element in the array. The angular
spectrum (AS) algorithm is used to efficiently compute the desired amplitude and phase
distributions for each desired pattern. Studies on animal behavior are conducted to
demonstrate that blind rats exhibit a visual-like behavioral response to ultrasound retinal
stimulation. Neuronal activities during ultrasound stimulation are recorded from the
contralateral superior colliculus surface using a 56-ch multi-electrode array. More detailed
and advanced safety studies have been conducted to evaluate the safety of US retinal
stimulation, including immunostaining, fundus and OCT imaging of the retina, and ERG
measurements. Last but not the least, three single-element transducers (center frequencies:
3.1, 5.4, and 20 MHz) are fabricated and used to examine the frequency-related stimulation
efficiency. The results provide in vivo proof that ARF is the physical mechanism of US
retinal stimulation.
Chapter 4 examines the feasibility of transcranial focused US (tFUS) stimulation on the
visual cortex to evoke neural activities. It is demonstrated that 0.5-MHz tFUS successfully

10

evokes neural activities in the visual cortex of both normal and retinal degenerate blind rats.
Three types of ultrasound waveforms were used in the three experimental groups. Various
types of cortical activity were observed when various US waveforms were used. In all rats,
when stimulated using continuous US waves, only short-duration responses were detected
at ultrasound “On” and “Off” time points. In comparison, pulsed waves evoked longer low-
frequency responses. Testing of various parameters of pulsed waves indicated that a pulse
repetition frequency higher than 100 Hz is needed to obtain low-frequency responses.
Based on the observed cortical activities, it was inferred that acoustic radiation force is the
predominant physical mechanism of ultrasound neuromodulation.
Chapter 5 summarizes this work on ultrasound stimulation for noninvasive visual
prostheses and the physical mechanism of ultrasound stimulation. It further addresses the
three aspects of future work: ultrasound array-based retinal stimulation with whole SC
recording, improved US VC stimulation, and the study of the biophysical mechanism of
US stimulation.

11

Chapter 2 Non-invasive Ultrasound Retinal Stimulation for
Vision Restoration
2.1 Introduction
Retinal degenerative (RD) disease, caused by progressive degeneration of the light-
sensitive photoreceptors in the retina, is one of the major causes of vision loss and
blindness worldwide. Despite the loss of sensitivity to light, the remainder of the visual
pathway is frequently intact and functional, allowing visual prostheses to emerge as
tools to restore visual function. Microelectronic retinal prosthetics were first proposed
in 1956 [56], which attempt to restore vision by bypassing the damaged photoreceptors
and stimulating directly in the inner retinal neurons. In recent years, some of these
devices have been translated from the laboratory to the clinic and implanted in patients,
such as Argus II [26, 57]. However, the challenges of the invasive implant of electronic
devices, limited amounts of electrodes, considerable surgical costs, and potential surgery
side-effects have remained unsolved. To conquer these challenges, many efforts have
been made, including the investigation of optogenetics [58, 59], near-infrared sensors
[47], chemical stimulation [60], and gene therapy [59, 61]. However, these methods still
require risky invasive procedures or comprehensive gene engineering.
Ultrasound (US) has been widely used in the clinic as a non-invasive approach for
diagnostic imaging and therapeutic applications. In recent years, with the attractive
features of noninvasiveness, deep penetration to cover the whole brain, and spatial
selectivity on the order of sub-millimeters, US neuromodulation has become a promising
technology in the field of neuroscience. Despite the mechanism of US neuromodulation
being still under investigation [62, 63], efforts have been made to exploit the potential
of US neuromodulation to treat various nerve-related diseases [64, 65].In terms of the
potential application of vision restoration, several pioneer studies have explored that the
US stimulation on the retina can evoke neuron activities [53, 66-68]. However, the lack
of an in vivo demonstration of vision restoration from degenerative retina impeded the
efficacy of US neuromodulation as a vision restoration approach.

12

In this work, we demonstrated that ultrasound stimulation on the retina could evoke
neuron activities in the contralateral visual pathways (superior colliculus and visual
cortex) in the brain. Especially in RD rats, our results first showed that the US could
reliably activate the degenerative retina in vivo while light stimulation has a silent
response. We further exploited the potential of US-based non-invasive visual prosthesis
by investigating the spatiotemporal resolution and safety of the proposed stimulation
method.
Before the stimulation scheme, the functionality of the recording system and the
vision of rats were examined using light stimulation. In our study, the light stimulation
response was recorded by the MEA from either the SC region (Fig. 2-4 a,c) or the VC
region (Fig. 2-5 a,c). In general, the normal-sighted rats had neuron reactions to light
stimulation, while the Royal College of Surgeon (RCS) RD rats did not have any
Figure 2-1. Schematic diagram of the experiment setup. (a) The retina stimulation part. A
focused single-element transducer was used to generate ultrasound waves targeting the retina.
Retinal neurons were excited and generated neural signals transmitting through optic nerve
to the brain. (b) The brain recording part. A multielectrode array (MEA) was inserted to the
contralateral superior colliculus (SC) or visual cortex (VC).

13

response, in accordance with our expectations. In SC recording, several MEA channels
had strong background noise, like channels 2 and 7 in Fig.2-4a. The reason should be
that the corresponding electrodes were placed in the blood vessels. Similar noise with
larger amplitude also happened in VC recording, and these channels were manually
deleted.

2.2 Methods
2.2.1 Experimental Setup
The schematic diagram of the experimental setup with ultrasound stimulation is
shown in Fig. 2-1. A dual-channel function generator (AFG3252C, Tektronix, Beaverton,
OR, USA) was implemented in this study to control the stimulus sequence and trigger
signals for data acquisition. More specifically, the output of Channel 1 was used to
generate the stimulus sequence, followed by an RF power amplifier (100A250A,
Amplifier Research, Souderton, PA, USA) with a gain of 50 dB, and then used to drive
a custom-built US transducer. Channel 2, which is self-synchronized with Channel 1
through the internal system clock, will send synchronized trigger signals to the interface
board of an electrode recording system - Lablynx (Neuralynx, Bozeman, MT, USA).
Regarding the light stimulation, a full-field strobe flash using a Grass Photic stimulator
(Grass Instrument Co., W. Warwick, RI, USA) was delivered to the contralateral eye in
the meanwhile, the stimulator sends out a trigger signal to the Lablynx recording system
for data recording. The time interval between each adjacent trigger signal is set to 6
seconds to ensure the evoked potential activities are back to normal.
During the experiment, the US transducer that was placed in front of the rat eye and
coupled with de-gassed gel was driven by a power-amplified sinusoid tone burst signal
controlled by the function generator. For electrophysiological signal recording from the
contralateral visual pathways, the contralateral side skull was first removed. Then, the
32-channel multi-channel electrode array (MEA) was either applied at the top layer of
the superior colliculus (SC) after removing the partial visual cortex (VC) for SC
recording or placed directly at the surface of VC for VC recording. In control groups,
the transducer stimulated surrounding areas around the eyeball.

14

To record the multi-unit neuron activities (MUA) from SC, a 4-by-8 32-channel high
impedance (that is 0.5 MOhms in this study) micro-electrode array (MEA, Microprobes
for Life Science, Gaithersburg, MD, USA) with a 150-um spacing between adjacent
electrodes were advanced into SC. The signals from MEA were sampled by the analog-
to-digital multiplexing headstage (HS-32-MUX-PTB, Neuralynx) before transferring to
the Lablynx recording system. VC region above the SC was removed to eliminate the
signal contamination from VC when doing the SC recording. The ultrasound transducer
and the recording system were grounded together to minimize the artifacts.
2.2.2 Animal Preparation
The in vivo rat experiment was performed according to the University of Southern
California Institutional Animal Care and Use committee (IACUC) protocol. A total of
twenty rats (male and around six-month-old) were investigated, including ten normally
sighted Long-Evan (LE) rats and ten RD Royal College of Surgeon (RCS) blind rats.
For either strain, eight rats were used for SC recording, and two rats were used for VC
recording. The RCS rats are characterized by retinal pigment epithelium (RPE)
dysfunction owing to the deletion of the Mer tyrosine kinase (MerTK) receptor that
abolishes the internalization of photoreceptor outer segments by RPE cells. All normal
LE rats were randomly and equally assigned to the studies of light stimulation or US
stimulation, and all RCS blind rats were also randomly and equally distributed to these
two stimulation modes, resulting in five rats in each group. For each rat, only one eye
was used for stimulation purposes, while another eye was untreated.
Before the experiment, the rats were anesthetized initially with an intraperitoneal
injection of Ketamine/Xylazine (50-90 mg/kg, 5-10mg/kg) and then with sevoflurane
inhalation through a nose cone. For electrophysiological signal recording, a tiny cranial
hole was made to expose the surface region of the VC and the corresponding SC
underneath. To ensure the sensitivity of the retina to light, all procedures were performed
in a dark room illuminated with dim red light. During the experiment, the rat eye was
first stimulated with light to establish the baseline of the retinal response, and then the
de-gassed ultrasound gel was used to couple the space between the transducer surface
and rat eye, finally tested with ultrasound stimulation to investigate its potential benefits.

15

After the experiment, the rats were sacrificed, and both eyes (i.e., the stimulated eye and
the untreated eye) were kept for histology analysis to investigate the safety issue of our
ultrasound stimulation sequence.
2.2.3 Ultrasound Transducer
Considering the size and the potential US attenuation of the eye, we have designed and
fabricated a 3.1 MHz transducer with a focal length of 10 mm and an f-number of 1. The
DL-47 (Del-Piezo Specialties, FL, USA) material was used as the piezoelectric layer due
to its high power-sustaining capability. A layer of 10-um parylene C was coated on the
surface of the transducer for protection and insulation. During the experiment, the
transducer was mounted on a 5-axis precision stage to accurately control the position of
the transducer. The acoustic fields of the transducer were calibrated using a hydrophone
(HGL-0085, ONDA Co, Sunnyvale, CA, USA). Two parameters were calculated,
including spatial peak pulse average intensity (ISPPA)and mechanical index (MI).
2.2.4 Ex-vivo Study
Two eyeballs were extracted and used to estimate the US attenuation in the eyeball
and the US-induced temperature increase. In attenuation measurement, the hydrophone
was placed at the focus of the transducer (10 mm away). The intact eyeball was held by
a clip and placed on top of the hydrophone with a gap of about 1 mm. The acoustic
pressure was measured with and without the presence of an eyeball. Measurements were
conducted three times for either eyeball. In temperature measurement, a needle-type
thermocouple was inserted into the posterior eye. A thermocouple meter was used to
record the temperature changes. The sclera was attached to the eyeball, which possibly
contributed to the higher measured attenuation than the simulated attenuation.
2.2.5 Stimulation Parameters and Patterns
To understand the effect of US stimulation parameters on the evoked neuron response,
two major stimulation parameters were investigated, including acoustic amplitude and
acoustic duration. First, we changed the ultrasound amplitudes (that is, the output
voltage of the function generator before feeding into the power amplifier) from 50 mv
to 500 mv with an interval of 50 mv. Then, we tested stimulation duration from 50 ms
to 200 ms with an interval of 50 ms. To compare the continuous tone burst mode with

16

pulse mode, we designed four different stimulation sequences, including one for
continuous tone burst mode and three for pulse mode with 30%, 50%, and 70% duty
cycles, respectively.
2.2.6 Data Processing
The recorded raw signals were sampled at 30 kHz and filtered with a 500-7000 Hz
digital filter to obtain neuron spikes. The recorded signals from VC were filtered with a
300-7000 Hz digital filter. For each channel, the maximal peaks with an amplitude three
times stronger than the background noise of this channel are considered spikes. The
response was considered to be observed when the average spike counts per 5 ms were
larger than 2. The response duration was determined by the time slot where the average
spike counts per 5 ms were continuously larger than 2. Signal processing was conducted
by MATLAB 2019b (MathWorks). The interpolation method used in response
distribution maps was 4-times modified Akima cubit 2-D interpolation. Since the
response maps at SC didn’t have a regular shape. An algorithm was used to
automatically determine its spatial resolution. Using the maximum point (value = 1 in
normalized figures) as the center, the half-amplitude beam widths in both x- and y-
directions were calculated and then averaged. The resolution was defined as half of the
averaged half-amplitude beam width.
2.2.7 Histology Analysis
At the end of the recording session, animals were killed with an overdose of halothane,
eyes were either immersed in Bouin’s fixative and embedded in paraffin or fixed in 4%
paraformaldehyde in 0·1 M Na-phosphate buffer, infiltrated with sucrose, and frozen in
Tissue Tek on dry ice. Transverse sections of the retina were cut, mounted on to slides,
and stained with hematoxylin-eosin (H&E). A series of sections through the full extent
of each transplant was evaluated at the light microscopic level. S-antigen
immunoreactivity was used to identify the residual photoreceptors in the host retina as
well as photoreceptors in the transplant. Briefly, deparaffinized sections were washed
with phosphate-buffered saline and incubated for 30 min in 20% horse serum. The
sections were incubated with a mouse monoclonal antibody against S-antigen (clone
A9C6) (Donoso et al., 1985) at a dilution of 1:20 000 overnight at 48C, and the binding

17

of the primary antibody was detected using the Vector Elite ABC kit for mouse
antibodies (Vector Laboratories, Burlingame, CA, USA). The antibody A9C6 is specific
for rods and blue cones (Donoso et al., 1985). Alternatively, frozen sections were
blocked in 20% goat serum and incubated with a 1:1000 dilution of A9C6 overnight.
After washing, the slides were incubated in a 1:100 dilution of Rhodamine Xanti mouse
IgG (Molecular Probes, Eugene, OR, USA), and covered slipped with DAPI-containing
Vectashield mounting medium (Vector Labs) for fluorescence. Sections were analyzed
using a Zeiss confocal microscope.
2.2.8 Finite Element Analysis Simulation
The FEA simulation was conducted by COMSOL Multiphysics 5.3a (Stockholm,
Sweden). An acoustic module and Bioheat transfer module were used. In the simulation,
the eyeball was simplified into four main parts: cornea, lens, vitreous body, and retina.
The shape and size of each part were set according to the [69, 70]. The acoustic and
thermodynamic properties of each part were set based on the prior study [71].
Parameters are listed in Table 2-1.

2.2.9 Statistical Analysis
Statistical significances between three or more were tested using ordinary one-way
ANOVA and Tukey’s multiple comparison test. Prism 9 software (GraphPad) was

DENSITY
(KG/M3)
SOUND
SPEED
(M/S)
HEAT
CAPACITY
AT
CONSTANT
PRESSURE
(J/KG/K)
THERMAL
CONDUCTIVITY
(W/M/K)
ATTENUATION
(DB/CM/MHZ)
WATER 1000 1500 4178 0.62 0
CORNEA 1062 1586 4178 0.58 0.78
VITREOUS 1005 1532 3999 0.6 0.01
LENS 1076 1647 3000 0.40 1.19
RETINA 1034 1538 3680 0.57 1.15
Table 2-1. List of acoustic and thermal parameters of water and organs of eye.

18

used to calculate the values. Significance values are p < 0.05 (*), p < 0.01 (**), p <
0.001 (***) and p < 0.0001 (****).

2.3 Results
2.3.1 Ultrasound Transducer and Power Calibration
A schematic overview of the US stimulation system and its implementation is shown in
Fig. 2-1. The custom-built US transducer has a central frequency of 3.1 MHz and a focal
depth of 10 mm. The full-width-half-maximum (FWHM) beamwidth and the depth of
focus (DOF) of the measured US beam are 640 µm and 4200 µm, respectively, resulting
in a good spatial resolution on the target retina tissue. The relationship between the output
Input Voltage (mV) Pressure (MPa) Isppa(W/cm^2) MI
50 0.432 6.220 0.245
70 0.535 9.535 0.304
100 0.699 16.306 0.397
150 0.943 29.639 0.536
200 1.289 55.416 0.732
250 1.481 73.121 0.841
300 1.742 101.144 0.989
350 2.030 137.353 1.153
400 2.283 173.790 1.297
450 2.530 213.398 1.437
500 2.825 265.979 1.604
550 3.086 317.368 1.753
600 3.374 379.374 1.916
Table 2-2. The relationship between the input voltage (before 50 dB gain) and ultrasound
pressure, intensity and mechanical index (MI).

19

voltage of the function generator and the free-field negative peak pressure at the focal point
and the 3D ultrasound field distribution was measured by the hydrophone and is shown in
Fig. 2-2.
To investigate the attenuation of US waves caused by the structures of the eyeball, finite
element analysis, and ex-vivo measurement were conducted to estimate the ultrasound
attenuation and US-induced temperature increase. As shown in Fig. 2-3, the eyeball
generated -2.0 dB attenuation and slightly enlarged the FWHM and DOF. The US-induced
(200 ms, NPP in free field = 2.28 MPa) temperature increase is less than 2 ℃. In ex-vivo
measurement, the attenuation caused by the eyeball was -3.3 ±0.4 dB. The US-induced
temperature increase is less than 0.6 ℃.

2.3.2 US-evoked Neuron Response in Rat in vivo
Prior to the stimulation scheme, the functionality of the recording system and the vision
of rats were examined using light stimulation. In our study, the light stimulation response
was recorded by the MEA from either the SC region (Figs. 2-4 ac) or the VC region (Figs.
2-5 ac). In general, the normal-sighted rats had neuron reactions to light stimulation, while
the Royal College of Surgeon (RCS) RD rats did not have any response, in accordance

Figure 2-2. The simulated acoustic field: (a) X-Z plane and (b) X-Y plane. (c) Measured
amplitude of pressure and MI.

20

with our expectations. In SC recording, several MEA channels had strong background
noise, like channels 2 and 7 in Figs. 2-4 ab. The reason should be that the corresponding
electrodes were placed in the blood vessels. Similar noise with larger amplitude also
happened in VC recording, and these channels were manually deleted.
We have shown examples of the US response collected from both the SC (Fig. 2-4 b)
and VC (Fig. 2-5 b) in the normal-sighted rat. Different from the light-evoked spike
activities recorded in most of the recording channels, the US response was only observed
at limited channels. This is because the light stimulator is a full-field strobe flash, which
will simultaneously expose a large area of posterior retina tissue, resulting in neuron
activities for most regions of SC and VC. By contrast, the US-stimulated region of interest
(ROI) determined by the focal zone of the US transducer is relatively as small as one point.
In other words, as shown in Fig. 2-2b, only a small number of channels that corresponded
to the stimulated retinal region in the retinotopic map were able to receive the spike
activities. A total of ten normal-sighted rats were used in this study to demonstrate whether
US stimulation is an efficient tool in evoking the neuron activities of the rats in vivo. SC
recordings were conducted on eight rats, and VC recordings were conducted on two rats.

Figure 2-3. The simulated acoustic field: (a)&(b) X-Z plane and (d) X-Y plane. (c)
Simulated temperature increase. (e) The axial acoustic pressure distribution along z-axis. (f)
The curve of temperature changes over time.

21

Although the magnitude variance existed among different rats, all of them had strong spike
Figure 2-4. Representative US-evoked neuron responses recorded from SC. (a) Light
response from normal sighted rat. (b) US response from normal sighted rat. (c) Light response
from blind rat. (d) US response from blind rat. The first row shows filtered signals by bandpass
filter between 500 Hz and 7000 Hz. The second row shows the average spike counts per 5
ms of all channels. In the third row, a representative channel was selected to demonstrate the
spike counts curve. The deep blue shows the average value and the baby blue shows the
standard deviation.

22

activities evoked by US stimulation.
The results in normal-sighted rats indicated that the US could be an alternative approach
to stimulate the retina in vivo. To further validate the advantageous use of US in vision
restoration, the RCS blind rat model with degenerated photoreceptors was used in this
study. Based on the results from ten RCS blind rats (Eight for SC recording and two for
VC recording), we demonstrated that US is still able to stimulate the remaining parts of the

Figure 2-5. Representative US-evoked neuron responses recorded from VC. (a) Light response
from normal sighted rat. (b) US response from normal sighted rat. (c) Light response from blind
rat. (d) US response from blind rat. The first row shows filtered signals by bandpass filter between
500 Hz and 7000 Hz. The second row shows the average spike counts per 5 ms of all channels.
In the third row, a representative channel was selected to demonstrate the spike counts curve.
The deep blue shows the average values, and the baby blue shows the standard deviation.

23

retina, followed by translating the electrical neural signals to evoke the neuron activities at
the SC (Fig. 2-4 d) and VC (Fig. 2-5d).
2.3.3 US Parameters on Response Pattern
Two major US stimulation parameters that have a substantial influence on evoked
response patterns were investigated, namely acoustic pressure and acoustic duration. To
conform to the U.S Food and Drug Administration (FDA) recommendation on the
maximum mechanical index (MI) of 1.9, which corresponds to a negative peak pressure

Figure 2-6. US-evoked neural responses change along with the US amplitude (a) and duration
(b). The statistical analysis of effect of US parameters. (c) Response intensity changes along
US duration. (d) Response duration changes along US duration. (e) Response intensity changes
along US amplitude. (f) Response duration changes along US amplitude. (g) The comparison
of responses between normal rats and retina-degenerated rats.

24

(NPP) of 3.35 MPa at 3.1 MHz [72], we varied the output voltage amplitude of the arbitrary
function generator from 200 mV (that is 1.29 MPa in NPP) to 600 mV (that is 3.37 MPa in
NPP) with an interval of 50 mV. Supplementary Fig. 1 shows the relationship between the
output voltage of the arbitrary function generator and the NPP at the focal point. Fig. 2-6a
shows the spike counts under various acoustic amplitudes. In addition to acoustic amplitude,
the acoustic duration from 5 ms to 200 ms was explored, in which the results are shown in
Fig. 2-6b. Fig. 2-6c-f lists the statistical comparisons among various stimulation amplitude
and duration on neuron response, such as spike count and response duration. All recordings
were repeated 12 times for statistical analysis. The intensity and duration of the responses
were investigated. The peak of averaged spike counts per 5 ms was used to indicate the
response intensity.
It was demonstrated that the response intensity did not change along with the US
duration when the duration was 10 ms or longer, while the response duration increased
according to the increase of US duration (Figs. 2-6c&d). As the acoustic pressure increased,
both response intensity and response duration increased (Figs. 2-6e&f). The overall
acoustic power exposure determined the threshold of evoking the neuron activity. As
shown in Fig. 2-6(g), either the combination of higher acoustic pressure and shorter
acoustic duration or the reversing combination was able to generate spike responses.
Comparing the neural responses from blind rats and normal-sighted rats, it was concluded
that, in general, the RCS blind rats have a higher threshold than normal-sighted rats.
In addition, we observed that both continuous tones burst mode and pulse mode with
duty cycle have the same capability to stimulate the retina in vivo. As indicated in Fig. 2-
7d, when the accumulated acoustic energy is the same in each US stimulation period (that
is, 6s interval between each adjacent US stimulation), the total number of spikes is the
same. However, a higher duty cycle tended to generate a larger response intensity (Fig. 2-
7c), which infers that the higher acoustic power led to higher neuron firing rates.

2.3.4 Spatiotemporal Resolution and Frame Rate of US stimulation

25

We further investigated the spatiotemporal resolution of US retina stimulation. The
response distribution maps at SC [73] were plotted by interpolating the response
amplitudes of all channels in MEA. The response amplitude of each channel is defined by
the total average spike counts in 500 ms. As shown in Fig. 2-8, the spatial resolution at SC
of US retina stimulation is smaller than 300 um. By manually moving the ultrasound
transducer along different directions, the major responding region in SC shifted
accordingly, which is demonstrated in Fig. 2-8c.
The possible frame rate of the US-based retinal prosthesis was also investigated by
decreasing both the interval between stimulations and the stimulation period until the

Figure 2-7. The effect of pulsed ultrasound stimulation. (a) A schematic diagram of pulsed US
waveform. (b) Representative responses with different duty cycles. (c) Statistical analysis of
the effect of duty cycle.

26

unattenuated neuron responses disappeared. A frame rate of 5 Hz was successfully
achieved using at least 5-ms stimulation with an acoustic amplitude of 2.53 MPa.
Stimulations were conducted for 40 seconds (200 times), and repetitive neuron responses

Figure 2-8. (a) Schematic diagram of the surface of SC and the MEA. (b) A representative
response mapping to show the resolution of response. (c) Different positions of SC were
activated by stimulating different areas. (d) Repetitive responses from stimulations with a PRF
of 5 Hz. (e) Latency difference between light response and US responses from normal sighted
rats and retina degenerated rats.

27

were observed after each stimulation without attenuation. A representative 2-second signal
fragment was shown in Fig. 2-8d.
We further investigated the response latency at SC between the light response and US
response on the normal-sighted and RCS blind rats. As shown in Fig. 2-8e, the onset
latency of the light response (20.18±31.77 ms) is significantly shorter than the US response
(43.88±9.55 ms), with a P-value < 0.001. The latency of RCS blind rats was significantly
longer than that of normal-sighted rats (P < 0.001), reaching 86.68±18.98 ms.

2.4 Discussion and Conclusion
In the past years, significant attempts have been made to restore a functional form of
vision to blind patients with visual prostheses. Although several types of visual prostheses,
including the retinal, optic nerve, and cortical prostheses, have been proposed, retinal
implants attracted the most promising attention since they can benefit from natural
information processing along the visual pathway. However, the current development of
retinal prostheses has focused on light-driven photovoltaic implants, which require
invasive surgery for implant devices. In addition, current technologies of electrical
stimulation have limited spatial resolution due to the number of stimulating electrodes, as
well as the inherent difficulty of complex implantation procedures of a large number of
electrodes.
As an attractive strategy, we successfully demonstrated that extraocular ultrasound
stimulation is able to elicit neuron activity in retinal degenerated rats in vivo with a high
spatiotemporal resolution and frame rate. It has been established that higher US frequency
is associated with better spatial resolution, resulting in a more precise manner for retinal
stimulation. However, the significant US attenuation and its potential thermal effect are
unavoidable at higher US frequencies, especially when cornea and lens tissue are presented
for in vivo study [71]. As suggested by Naor et al. [66], the center frequency of the US
transducer ranging from 2 to 10 MHz, could provide spatial resolution similar to that of the
first FDA-approved retinal prosthesis – Argus II. Therefore, to achieve in vivo feasibility,
good spatial resolution, and the capability of sustaining high power and long duration, a
3.1-MHz US transducer was designed and fabricated in this study.

28

Owing that visual information travels along the optic nerves, is pre-processed in sub-
cortical relays such as the SC (receives projections from 85% to 90% of the RGCs directly),
and is finally processed in the VC, our experiment was designed to record neuron responses
from both SC and VC via the multi-electrode array directly. To ensure a fair comparison
with natural light stimulation that is transferred to bioelectrical signals by photoreceptors,
the US stimulation was first conducted in normal-sighted rats in vivo. As expected, we
observed similar evoked neuron activities in SC and VC under both light and US
stimulation, respectively. Such results demonstrated that the US could be used as an
alternative approach to generating encoded information expressed as action-potential
sequences.
To further illustrate the advantages of using US for degenerative blindness, the
effectiveness of US stimulation was investigated in retinal degenerate RCS rats in vivo.
With the loss of functional photoreceptors, light cannot evoke neuron activities at either
SC or VC of RCS rats in vivo. By comparison, US stimulation still partially activated
interneurons beyond photoreceptors [53]. Neuron responses were still observed, while the
magnitude of the response was relatively weaker than that of normally sighted rats under
the same US parameters. Overall, our results are the first time to discern the neuron
activities in the retinal degenerate RCS rats in vivo. Although the functional mechanism,
such as which parts of the retina are activated by the US, was still under investigation,
utilizing the US as a non-invasive stimulation of retinal neural prostheses paved the way
for the future application of ultrasound retinal treatment to some extent.
For either in vivo study or future translational study, the acoustic energy exposure of
using the US as a neuromodulation approach should be considered to ensure safety.
However, there is no standard regulation such as intensity level or mechanical index (MI)
to follow due to the newly developed technology. Therefore, we demonstrate that our US
stimulation approach is safe from two aspects to some extent. First, we measured acoustic
pressure via hydrophone and then calculated the MI under all experimental parameters
setup we used for the simulation study. Compared with previous US stimulation studies,
our implemented acoustic parameters were within a reasonable range. Second, to confirm
whether US stimulation has the potential to damage retinal tissue, we collected a

29

histological image of the stimulated retinal sample (Fig. 2-9). All results indicated that our
stimulation parameters did not have the potential to damage the retinal tissue.
The differences among latencies of light stimulation and US stimulation in normal-
sighted rats and RCS blind rats were observed in this study (Fig. 2-8e). This phenomenon
was also observed in electrical stimulation [74]. The reason is unclear, and further
investigations are required. The US-sensitive neurons could either respond slower than

Figure 2-9. Representative histology results of US stimulated retinas from both a normal-
sighted rat and a RCS blind rat.

30

light-sensitive photoreceptors or US-sensitive neurons need to accumulate the energy for
several milliseconds before activation. An ex-vivo study by Menz reported that the latency
of US stimulation decreases by more than 100 ms as the US power increase [68]. It is also
interesting to notice that the US stimulation latency of RD blind rats was longer than the
latency of normal-sighted rats. This may infer that the degenerated part in RD rats also
contributes to the neuron activation and signal transduction caused by the US.
The investigation of the ultrasound neuromodulation mechanism is twofold: the
physical mechanism and the neurological mechanism. Ultrasound is a kind of mechanical
wave with three main effects, acoustic radiation force (ARF), cavitation, and thermal effect.
Menz et al. have concluded that the acoustic radiation force is the main effect of ex-vivo
ultrasound retina stimulation by carefully investigating the relationships between neuron
activities and ultrasound parameters, including frequency, intensity, and standing waves.
Although the conditions for an in-vivo study could be different, our results tend to support
the ARF as well. Given the ultrasound parameters of 2.28 MPa and 200 ms, the ultrasound-
induced temperature increase is less than 2 ℃, which is unlikely to evoke temperature-
sensitive neurons [75]. The temperature increase in in-vivo conditions should be even lower
due to the blood flow perfusion. MI is usually used to indicate the ultrasonic cavitation-
related bioeffects. Since the MI used in this study is lower than the FDA-required threshold
of 1.9, the cavitation effect is not considered the main reason.
The neurological mechanism of ultrasound retina stimulation is still inconclusive. The
well-known pressure phosphene phenomenon that applying mechanical pressure on
eyeballs can generate illusory light spots when there is no light inspired the idea of
ultrasound retina stimulation [76, 77]. It is further demonstrated mechanosensitive ion
channels in retinal neurons are responsive to localized mechanical stimulation [78, 79].
However, many types of mechanosensitive ion channels have been reported to be
functional in retinal cells, including TWIK-related K+, transient receptor potential (TRP)
channels, Piezo 1, and others [79-81]. The role of these channels in ultrasound stimulation
is unknown. Some studies have investigated the roles of different types of
mechanosensitive ion channels in mechanical and ultrasound stimulation. However, the
results are not consistent for now. For example, TRP vanilloid 4 has been proven to be
important in mechanical and ultrasound stimulation [78, 79, 82], while it was also reported

31

that TRP polycystic 1/2, TRP canonical 1, and Piezo 1 are important in ultrasound
stimulation, instead of TRP vanilloid channels [80]. Since the stimulation methods were
different in these studies, a possible explanation for the inconsistency is that different types
of mechanosensitive ion channels response to different mechanical stimulations.
Another interesting phenomenon of ultrasound neuron stimulation is its relationship to
the PRF. Our results demonstrated that continuous ultrasound waves and pulsed ultrasound
waves with different PRF generated similar neuron activities. However, the research
conducted by Kubanek et al. demonstrated that the behavior response of Caenorhabditis
elegans to ultrasound stimulation depends on the PRF and duty cycle. The ideal parameters
are around 1 kHz PRF and 50% duty cycle [83]. A possible hypothesis is that certain types
of neuron cells or mechanosensitive ion channels are selectively sensitive to the stimulation
with a certain frequency range, like the auditory nerve fibers and hair cells.
Some studies have shown that pulsed ultrasound stimulation can activate the auditory
pathway [84], and the response can further spread to other cortex areas [85]. It was also
reported that patients with earplugs heard tones in the same frequency as Doppler
ultrasound’s repetitive frequency [86]. These phenomena raised the concern that
ultrasound neuromodulation does not directly excite targeted neurons. The evoked neuron
activities could be side effects of the response initiated from the auditory pathway.
However, this concern does not apply to our study. The spatial mapping results also
demonstrated that the neuron responses are related to the stimulation position and are
highly confined to a small area. They all suggested that the responses at SC and VC were
the neural signals transmitting in the visual pathway, which originated from the US-
stimulated retina.
The next step of this study would use an ultrasonic 2-D array for retina stimulation.
Owing to the fixed focal zone of the single-element-based US transducer, it is difficult to
precisely alter the focal zone along the retina with curvature, which hinders the attempts to
accurately generate arbitrary patterns on the retina. The future development of a 2D matrix
array with the ability to electronically steer the beam in a 3D domain will be helpful in
pattern generation. Another factor that would affect the accuracy of the beam pattern is the
anatomy of the eyeball, especially the cornea and lens in the anterior eye. The reflection
and refraction of ultrasound waves caused by these parts can distort the designed beam

32

pattern. However, the array’s ability to independently control the amplitude and phase of
each element can compensate for the distortion using an inversion algorithm [87, 88]. In
conclusion, these results represent a step towards non-invasive retinal prosthesis through
the use of a US-based approach. Different from the standard light response, US responses
have various response patterns which are modulated by acoustic intensity and duration.
The in vivo demonstration on retinal-degenerated rats suggests that the US can reliably
stimulate and evoke neuromodulation, which opens a new avenue for the development of
non-invasive retinal neural prostheses.

33

Chapter 3 2D-array Based Ultrasonic Retinal Prosthesis and
Its Frequency Dependent Efficiency
3.1 Introduction
Retinal degenerative (RD) disease, characterized by progressive degeneration of the
light-sensitive photoreceptors in the retina, is one of the major causes of vision loss and
blindness worldwide. Age-related macular degeneration (AMD) is a leading cause of retina
degeneration affecting 196 million people in the world [89]. Another cause is retinitis
pigmentosa which has a worldwide prevalence of 1:3000 to 1:7000 people [90]. Despite
the loss of sensitivity to light, the remainder of the visual pathway in retina-degeneration
patients is mostly intact and functional, allowing prostheses as a useful tool to restore lost
vision. The current treatment strategy for blindness caused by retina degeneration is to use
invasive retinal prostheses based on electric stimulation. Non-invasive ultrasound
stimulation on the retina has been considered a more promising technique for vision
restoration and has been investigated by several different groups both in vitro and in vivo
[53, 66-68]. Our work (Chapter 2 of this thesis) reported the first observation of ultrasound-
evoked visual signals from in-vivo rats that are blind due to retinal degeneration [91]. A
summary of ultrasound retinal stimulation is listed in Table 3-1.
With all prior validation and mechanism studies, it is time to investigate and
demonstrate a practical ultrasonic retinal prosthesis. So far, no ultrasonic retina prosthesis
that could dynamically stimulate the retina has been developed. Also, no animal behavior
responses have ever been observed, and long-term safety investigation is lacking. In this
chapter, we are going to demonstrate a 2D-array-based ultrasonic retinal prosthesis and
address all the pre-mentioned concerns. We also investigate the frequency-dependent
stimulation efficiency and provide a further understanding and in-vivo proof of the physical
mechanism of ultrasound retina stimulation.
3.2 Materials and Methods
3.2.1 Ultrasound 2D Array Design, Fabrication

34

A 16-by-16 2D ultrasonic array transducer with 4.5-MHz center frequency and 750-µm
(~2λ) pitch was designed and fabricated for this study. An illustration of the transducer
design and a photo of the fabricated 2D array are shown in Fig. 3-1. Since the array was
designed for excitation, the requirements of our array are different from the arrays used for
imaging. Piezoelectric materials with a high mechanical quality factor, low dielectric and
mechanical loss, high Curie temperature, and high coercive field are desired for high-power
ultrasound transducers. A narrow bandwidth would not be a problem for stimulation
transducers. In this study, a similar concept is applied to a fully electronically controlled 2-
D array. To improve the performance for single-frequency power generation and minimize
attenuation, matching layers were not used at the front face of the array. We used an
interposer filled with silver epoxy paste as a light-conducting backing and applied a PZT-
Interposer-PCB structure [92] for this array. Using the interposer as the backing relaxes the
required tolerance on the flatness of the surface of the printed circuit board (PCB) or
flexible printed circuit (FPC) for assembly.
Normally, an interposer backing using acoustically attenuating material will provide
significant attenuation of the acoustic wave, leading to reduced ringdown and a wider
bandwidth at the expense of reduced power transmission and sensitivity. This feature is
advantageous for an imaging array. However, not suitable for stimulation, in which our
main goal is to provide efficient power transmission at a single-design frequency. To match

Figure 3-1. (a) The layered structure of an element in the array. (b) The photo of the fabricated
array together with interposer and PCB board.

35

these requirements, we designed the interposer differently. To reduce the amount of
backside attenuation and thereby increase the resonance and output power of the elements,
the conducting interposer pillars were designed to be a smaller volume fraction of the
combined azimuthal and elevational pitch of the elements. In this way, the interposer is
mainly a frame of acrylic, which makes it a low-attenuation backing. The fabrication
process of the acoustic stack of the 2-D array transducer is similar to our previous work
[92]. Hard PZT material (DL-48, DeL Piezo Specialties, LLC, West Palm Beach, FL, USA)
was used for array fabrication. It had a pitch width of 750 μm and kerf width of 100-μm
diced by a high-speed dicing saw (Tcar 864-1, Thermocarbon, Casselberry, FL, USA). The
finished piezoelectric material was lapped down to a thickness of 350 μm to achieve the
designed center frequency (4.5 MHz). Then the sample was then mechanically polished
and sputtered with a Cr/Au (50/100 nm) electrode using a sputtering system (NSC-3000
Sputter Coater, Nano-Master, Inc., Austin, TX, USA). An interposer backing was
fabricated using the process described in detail in [93]. A 4-mm tall support grid with a
minimum wall thickness of 108 μm was 3-D printed in acrylic (3-D Systems, Rock Hill,
SC, USA) and then filled with a conducting silver epoxy paste (E-Solder 3022, Von Roll
Isola, New Haven, CT, USA). The E-solder forms pillars spaced at the array pitch (750 μm)
and electrically isolated by acrylic walls. The piezoelectric sample was then bonded to the
interposer backing using a thin layer of E-solder. After the assembly had cured, the
piezoelectric material was diced along both azimuthal and elevational directions to create
2-D array elements, resulting in a total of 16 × 16 2-D array elements. It had a pitch width
of 750 μm and kerf width of 100-μm diced by a high-speed dicing saw (Tcar 864-1,
Thermocarbon, Casselberry, FL, USA). The kerfs were filled with EPO-TEK 301 (Epoxy
Technology, Billerica, MA, USA) and allowed to cure. This was followed by sputtering a
Cr/Au (50/100 nm) electrode to create the top electrode linking all the elements as the
ground connection. The finished acoustic stack had 256 active elements in total. After the
fabrication was completed, the acoustic stack was bonded with the FPC using E-solder
3022. The FPC also comprised two connectors for interface with two modified Philips/ATL
L7-4 cables, which were then connected to the ultrasound system (Vantage 256, Verasonics,
Inc., Kirkland, WA, USA).

36

Table 3-1. Summary of Ultrasound Retinal Stimulation Results
Authors Naor et al. Menz et al. Menz et al. Jiang et al. Qian et al.
Species Sprague Dawley rat
(in vivo)
Isolated salamander
retinas (ex vivo)
Isolated salamander
retinas
(ex vivo)
Isolated Sprague-
Dawley rat retina (ex
vivo)
Long Evans rats and RCS
retina degeneration rats (in
vivo)
Frequency 0.5 & 1 MHz 43 MHz 0.5 – 43 MHz 2.25 MHz 3.1 & 4.4 MHz
Acoustic
Pressure
0.09-0.16 MPa at 0.5
MHz
0.56 – 0.72 MPa at 1
MHz
0.77 – 1.34 MPa 0.2 - 1.9 MPa 0.34 MPa 1.29 – 3.09 MPa
Stimulation
Duration
5 – 20 ms 30 – 1000 ms 5 – 1000 ms 400 ms 1 - 200 ms
Major
Outcomes and
Findings
The first study showing
ultrasound retinal
stimulation can safely
evoke visual signals in
vivo.
Results suggesting
that ultrasound
stimulation has the
potential to treat
blindness with
photoreceptor
degeneration diseases.
Acoustic radiation
force is the physical
mechanism of
ultrasound retinal
stimulation.
First observation of
cell-type specific
response from
ultrasound retinal
stimulation.
First demonstration of
visual signals evoked by
ultrasound retina stimulation
in blind rats and the
patterned ultrasound
stimulation.

37

3.2.2 Neuron-recording Experiment Setup
The schematic diagram of the experimental setup with ultrasound stimulation is shown
in Fig. 3-2. A function generator (AFG3252C, Tektronix, Beaverton, OR, USA) was
implemented in this study to synchronize the ultrasound stimulation system (Vantage 256,
Verasonics, Inc., Kirkland, WA, USA) and multi-channel electrophysiological data
acquisition Lablynx system (Neuralynx, Bozeman, MT, USA). In light stimulation, a full-
field strobe flash using a Grass Photic stimulator (Grass Instrument Co., W. Warwick, RI,
USA) was delivered to the contralateral eye. In the meanwhile, the stimulator sends out a
trigger signal to the Lablynx recording system for data recording. The time interval
between each adjacent trigger signal is set to 6 seconds in order to ensure the evoked
potential activities are back to normal.
Figure 3-2. Schematic diagram of ultrasound retina stimulation system. Top: The workflow of
2D-array ultrasound retina stimulation with dynamic patterns. Calculated amplitude and phase
distribution are applied to 2D array. Then the generated patterns are used to stimulate retina in
vivo. MEA was used to map the neuron responses from the brain. Bot: Verasonics ultrasound
system. The detailed structure of one element in the array. The ultrasound evoked neuron
responses with a “C” pattern.

38

During the experiment, the US 2D array was connected to a 3D-printed collimator,
which controls the distance between the array and eyeball to be the exact 10 mm. The
collimator was filled with de-gassed ultrasound gel. The 2D array was controlled by the
Verasonics Ultrasonic system, which can control the amplitude and phase of each element
in the 2D array. By applying the correct amplitude and phase distribution, an ultrasound
2D array can generate arbitrary desired patterns. The algorithm to calculate the amplitude
and phase distribution will be elaborated in the following algorithm section (3.2.4).
For electrophysiological signal recording from the contralateral visual pathways, the
contralateral side skull was first removed. Then, the 56-channel multi-channel electrode
array (MEA) was applied at the top layer of the superior colliculus (SC) after removing the
partial visual cortex (VC) for SC recording. Signals from MEA were sampled by two
analog-to-digital multiplexing headstages (HS-32-MUX-PTB, Neuralynx) before
transferring to the Lablynx recording system. VC region above the SC was removed in
order to eliminate the signal contamination from VC when doing the SC recording. The
ultrasound transducer and the recording system were grounded together to minimize the
artifacts. Different from the 32-channel MEA used in the prior study, this 56-channel 0.5-
MOhms MEA (Microprobes for Life Science, Gaithersburg, MD, USA) was specially
designed for SC mapping with a 350-µm spacing between adjacent electrodes. The
electrode distribution shape is shown in Fig. 3-3. Because the surface of SC is curved, red

Figure 3-3. The design of the new MEA to cover the whole surface of the SC.

39

electrodes are 100 µm longer than blue electrodes to ensure all electrodes can touch the
surface of SC at the same time. The recorded MUA signals were filtered (300 - 7500 Hz),
rectified, then filtered again (0-300 Hz). The amplitudes of the processed MUA signals are
used to plot the SC neuron response mapping.
3.2.3 Water-licking Behavioral Experiment Setup
Water-licking experiment setup includes four different parts. The first part is an animal
holder. Animals were fastened and held on a stage. A night vision camera targets the head
and mouth of the animal to record the water-licking behavior. The second part is a water
pump, which silently supplies water drops when it is triggered. The third part is an LED
flashlight. To ensure the light-evoked visual signals are comparable to focused ultrasound-
evoked visual signals, the LED light was covered by black paper, and only a small point

Figure 3-4. The schematic diagram of the water-licking behavioral experiment. The Arduino is
used to control and synchronize LED flashlight, water pump, and ultrasound stimulation. A night
vision camera is used to monitor the water-licking behavior of rats.

40

came out. The fourth part is the ultrasound stimulation system, which includes a function
generator, RF amplifier, and ultrasound probes. The computer-controlled Arduino is used
to control and synchronize all four parts.
3.2.4 Algorithm for Pattern Generation
It is an inverse wave propagation problem to calculate the desired amplitude and phase
distribution to generate arbitrary patterns. The basic idea is to use the desired pattern
(pattern plane) as the transmitter and propagate the wave to the ultrasound 2D array (array
plane). Due to the reciprocity of acoustic waves, when the achieved amplitude and phase
distribution in the array plane is applied to 2D arrays, the desired pattern will be reproduced.
While various inverse wave propagation algorithms are available, there is always a tradeoff
between computation efficiency and accuracy (the level of approximation in the
calculation). The band-limited angular spectrum method is used in our study [87] due to its
high computation efficiency and accuracy in our applications.
3.2.5 Animal Preparation
All animal procedures were approved by the University of Southern California
Institutional Animal Care and Use Committee (IACUC). Animals are divided into two
different groups: 1) the neuron-recording group and 2) the behavior-test group.
1) Neuron-recording group. Animal preparation and procedures in this group are
almost the same as in the prior study (Chapter 2). Both normally sighted Long-Evan
(LE) rats and RD Royal College of Surgeon (RCS) blind rats were used. The RCS
rats are characterized by retinal pigment epithelium (RPE) dysfunction due to the
deletion of the Mer tyrosine kinase (MerTK) receptor that abolishes the
internalization of photoreceptor (PR) outer segments by RPE cells. All rats were
male and about six-month-old. The rats were anesthetized initially with an
intraperitoneal injection of Ketamine/Xylazine (50-90 mg/kg, 5-10mg/kg) then with
sevoflurane inhalation through a nose cone. [94] The eyes were dilated using 1%
tropicamide and 2.5% phenylephrine drops. The cranium was exposed by removing
the skin above the skull. A small cranial hole was made using a dental drill. All
procedures and experiments were performed in a dark room illuminated with dim

41

red light to minimize possible stimulation of the VC due to photoreceptor activation.
The space between the brain surface and transducer was filled using US gel.
2) Water-licking behavior test group. Water-licking experiments require the animal's
head to remain immobile, so six cranial nails and a metal support rod were surgically
implanted into the skull of normal Long Evans (LE) rats to serve as fixed ports. The
implants were attached to this using a light-curing adhesive (Flow-It™ ALC™),
reinforced with dental cement, and wrapped with a flowable dental composite
(Kwik-Sil™) to ensure implant stability. The animals were allowed to recover for 3
days. There was a 2-day water restriction period before the training period when the
animals were given only a small amount of water to ensure that the rats performed
licking activity to obtain water. During the training phase, the rat was held in the
holder, their arms and legs were secured, and the nails on their head were attached
to the immobilizer to ensure that the head of the rat remained immobile. The rats
received 20 training sessions per day for 8 consecutive days. Each session consisted
of 0.5s light stimulation, followed by a 3s waiting period. After the waiting period,
two drops of water were provided through a water hose to establish light stimulation
as positive reinforcement, and then there was a 30 s waiting period for rats to rest.
The number of lickings performed by the rat at each session was counted. After the
completion of the 8-day training, light stimulation was replaced with ultrasound
stimulation (3.3MHz, 400mVpp). During this testing phase, the experimental set-up
remained consistent with the training phase, and the experiments were completed
after two days of testing.
3.2.6 Ultrasound Imaging Guide for Retinal Stimulation
The 4.5-MHz ultrasound 2D array is also used for 3D imaging of the eyeball and
locating the retina to guide the retinal stimulation. Ultrasound imaging is also conducted
using Verasonics systems. Plane-wave imaging [95] with 37 angles compounding in both
azimuthal and elevational directions is conducted to the 3D imaging of the eyeball before
the stimulation.
Inter-frame image processing is also conducted to retrieve the blood flow information
enhancing the contrast between the retina and sclera and providing a more accurate location

42

of the retina as well as the optic nerve head. Since the acoustic scatters in blood flow are
blood cells, whose local concentration is time-variant, the ultrasound signal intensity of
blood flow changes along the time, and the intensity change rate is proportional to the flow
speed. Therefore, the basic idea of ultrasound flow imaging is to do a high-pass inter-frame
filter of the acquired B-mode image sequence. In this way, the time-variant blood flow
signals can be retrieved and separated from signals of other tissues that do not move or
move at a much lower frequency. Detailed theory, validation, and method of ultrasound
flow imaging can be found in [96].

3.2.7 Single-element Transducers Design and Fabrication
In the frequency-dependent stimulation efficiency study, we designed and fabricated
three single-element transducers with different center frequencies. The first one is the prior
used 3.1 MHz transducer with a focal length of 10 mm and an f-number of 1. The DL-47
(Del-Piezo Specialties, FL, USA) material was used as the piezoelectric layer due to its

Figure 3-5. Top row: the hydrophone measured frequency responses of all three transducers. Bot
row: the 2D mapping of focus of each transducer at the focal plane. Higher-frequency transducers
have finer focal points. The FWHM is 640 µm of the 3.1-MHz transducer, 510 µm of the 5.4-
MHz transducer, and 110 µm of the 20-MHz transducer.

43

high power-sustaining capability. A layer of 10-um parylene was coated on the surface of
the transducer for protection and insulation. The second transducer is 5.4 MHz and made
of DL-47 as well. The third transducer is centered at 20 MHz. Due to the significant
difference in the frequency, it is made Lithium Niobate (LNO), taking advantage of LNO’s
stability under high voltage and temperature conditions. The acoustic fields of the
transducer were calibrated and mapped using a hydrophone (HGL-0085, ONDA Co,
Sunnyvale, CA, USA) and a 3D scanning system. The measured frequency response and
focus mapping are shown in Fig. 3-3. Higher-frequency transducers have finer focal points.
3.3 Results
3.3.1 Ultrasound Imaging-guided Retinal Stimulation
An imaging guide is an important and desired function for non-invasive stimulation
devices because the organ structures among individuals can be geometrically different and
time-variant. Also, the relative positions and directions between the stimulation device and

Figure 3-6. 4.5-MHz ultrasound imaging using the 2D array. Although the image quality is
relatively poor, eyeball structures still can be distinguished.

44

target organs could change because of the unexpected movement during long-time wear.
An intrinsic advantage of the ultrasonic prosthesis is that the US neurostimulator (US
transducers) also has the ability to do imaging of deep tissues. Although as we mentioned
before, the 2D array is specifically designed for power transmission instead of imaging, a
relatively low-quality but useful 3D imaging of the eyeball is still achievable using this 2D
array. Here, as shown in Fig. 3-6, we demonstrate that 3D imaging of the eyeball is
successfully achieved by using our 4.5-MHz 2D array.
3.3.2 Ultrasound Pattern Generation and Stimulation
By using the inverse algorithm (band-limited angular spectrum method), arbitrary
patterns can be generated using an ultrasound 2D array. Here we demonstrated that
different letters can be generated using our 256-ch 2D 4.5-MHz array. The pattern plane is
20 mm away from the surface of the array so they can be applied on the rat retina during
the experiment. The “C” pattern is used as a representative example. Fig. 3-7(a)&(b) show

Figure 3-7. (a) Amplitude distribution of 2D array for “C” pattern. (b) Phase distribution of 2D
array for “C” pattern. (c) Simulated generated “C” at the pattern plane (20 mm away from the
array surface). (d) Mapped neuron responses at SC of “C” pattern. (e) Mapped neuron responses
at SC of “V” pattern. (f) Mapped neuron responses at SC of “S” pattern.

45

the calculated amplitude and phase distribution of the 2D array to generate pattern “C”.

Figure 3-8. (a) The frequency spectrum measured by hydrophone of the single-element
transducers used for mechanism study. Transducers were drive at different frequencies and their
output pressures were calibrated with the spectrum. (b) Frequency-related pressure threshold. (c)
Frequency-related mechanical index threshold.

46

The generated pattern at a 20 mm distance is shown in Fig. 3-7(c). Then, this pattern is
used to stimulate the retina, and the neuron responses at SC were mapped using MEA.
Response mapping is shown in Fig. 3-7(d). Gray color points indicate the location of
electrodes. We further showed letters “V” and “S” are successfully achieved in rat brains.
3.3.3 Frequency-dependent Stimulation Efficiency
Safety and power consumption always are concerns of a long-term prosthesis device,
and they can be improved by using more efficient ultrasound stimulation parameters. For
example, in [53], Menz et al. conducted an ex-vivo study on an isolated salamander retina
and showed that the acoustic pressure threshold to activate retinal neurons decreases when
ultrasound center frequency increases. If this is also the case of in-vivo retinal stimulation,
then the power consumption and mechanical index of ultrasound can be significantly
reduced by using a higher-frequency ultrasound. On the other hand, investigation of
frequency-dependent stimulation efficiency will reveal the physical mechanism. Therefore,
we designed and fabricated three single-element transducers (center frequencies: 3.1, 5.4,
and 20 MHz) that cover the frequency range from 2 MHz to 22 MHz to investigate the
frequency-related stimulation efficiency. In vivo experiments were conducted on RCS
retina degenerated rats. Neuron activities during ultrasound stimulation were recorded
from the contralateral superior colliculus (SC) surface using a multi-electrode array (MEA).
The hydrophone-measured frequency responses and the focal spot of all three transducers
are shown in Fig. 3-5. Here the neuron responses from ultrasound at different frequencies
are shown in Fig. 3-8. Fig 3-8(a) shows the representative processed neuron responses. The
processing details are included in section 3.2.2. An average response amplitude larger than
25 µV within the time window of 20 ms – 200 ms (Ultrasound stimulation happens at time
0) is recognized as successfully activated retinal neuron responses. To determine the
pressure threshold at each ultrasound frequency, we change the input voltage to find the
lowest ultrasound pressure that can successfully activate retinal neurons. The acquired
results are shown in Fig. 3-8(b). Acoustic pressures in both free space and after eyeball
attenuation are shown. It can be noticed that a much higher free-space acoustic pressure is
required for active neurons in the high-frequency range, as shown in blue lines. However,
since the attenuation of high-frequency ultrasound is also much higher, the attenuated
acoustic pressure of high-frequency ultrasound on the retina is actually lower than the

47

attenuated lower-frequency ultrasound, as shown in the pink line. We further calculate the
mechanical index (MI) threshold, and the results are shown in Fig. 3-8(c). When the
frequency is higher than 4 MHz, the required MI is lower than 1.9, which is an FDA
requirement for diagnostic ultrasound in normal tissues. When the frequency is higher than
20 MHz, the required MI is lower than 0.23, which is an FDA requirement for diagnostic
ultrasound in the eye. Therefore, our results demonstrate that high-frequency ultrasound
retinal stimulation is safe and within the FDA requirements.
Another significance of this result is that it is the first in-vivo evidence that ARF is the
physical mechanism of ultrasound retinal stimulation. The cavitation effect is ruled out

Figure 3-9. Fundus imaging and OCT of retina (a) before and (b) after ultrasound stimulation.

48

because the cavitation effect is inverse proportional to the ultrasound frequency. Also, the
heating effect is negligible [53, 91] both in vitro and in vivo. Therefore, ARF is concluded
to be the only possible physical mechanism.
3.3.4 Safety Investigation
To further investigate the safety of ultrasound retinal prosthesis, we conducted
comprehensive examinations of the eyeball and retina before and after the stimulation. The

Figure 3-10. ERG results before and after ultrasound stimulation. (a)&(b) show the ERG
signals after two weeks of ultrasound stimulation using a normal intensity. (c)&(d) show the
ERG of the untreated eye (control group) before and after the stimulation. (e)&(f) are the
negative control, showing the ERG change of the eye stimulated by excessive ultrasound
intensity (two times stronger than the normal intensity).

49

examinations include 1) Fundus retina imaging, 2) OCT imaging of the retina, and 3)
Electroretinogram (ERG) measurement. In the end, animals were sacrificed, and eyeballs
were collected for histology. Then H&E staining, GFAP staining, and CD68 staining were
conducted to examine if there was any potential morphological and functional damage
caused by ultrasound stimulation.

Figure 3-11. Representative histology results of stimulated and control eyeball. (a) H&E staining,
(b) GFAP staining, (c) CD68 staining.

50

In this experiment group, three healthy rats were used. Imaging and ERG measurements
of both eyes were conducted first, followed by a two-week stimulation (one hour per day).
For each animal, one eye was selected for stimulation, and the eye was kept as the control
group. One week after the completion of stimulation, imaging and ERG measurements
were conducted again. Then animals were sacrificed, and both eyeballs were collected for
histology and staining analysis.
No damage is observed in all comprehensive examinations. Fundus imaging and OCT
imaging of the retina before and after stimulation are shown in Fig. 3-9(a) & (b). No
morphological changes or thickness change of the retina is observed. Also, the clear image
quality indicates the clearance of the whole eyeball, suggesting no blur, vitreous
hemorrhage, or cataract was induced by stimulation. Fig. 3-10 shows the measured ERG
signals before and after the stimulation. Fig. 3-9 shows the H&E staining, GFAP staining,
and CD68 staining results of both stimulated and controlled eyeballs. These results suggest
there is no functional and cell-level damage caused by ultrasound stimulation. In
conclusion, by conducting comprehensive examinations at different levels: from the whole
eyeball to cell, from morphology to function, all results indicate that ultrasound retinal
prosthesis is a safe technology for long-term stimulation.
3.3.5 Water-licking Behavioral Test
Apart from the electrophysiological recording results, behavioral tests can provide more
direct and conclusive evidence of the efficacy of ultrasound retinal prosthesis. There are
various behavioral tests available for vision ability, including spatial maze escape, electric
shock fear condition, and water-licking [97]. Water-licking behavior test [47] was selected
in this study because it minimized harm to subjects. Also, it can be performed on fastened
subjects, which makes the ultrasound stimulation system and transducer placement much
easier and more stable. Results of behavior tests of normal rats are shown in Fig. 3-12. Fig.
3-12(a) shows the licking action happens before, during, and after the stimulation applies.
Before the stimulation, the random licking rate is kept at a relatively low level (<2 Hz).
After the stimulation, the licking rate increased because rats learned the relationship and
expected water drops. After water drops, the licking rates reach their peak. Licking
responses to both light and ultrasound stimulation have similar curves. Fig. 3-12(b) shows

51

the anticipatory lick number in the 10 days of training. On the 8th day, the stimulation
method is switched from pointed light to ultrasound. There is no significant change in the
licking rate. Fig. 3-12(c) shows the raster plot of water-licking behavior among multiple
tests from both ultrasound and light stimulation. Fig. 3-12(d) shows the summary and
statistical comparison of licking rate in three different stages (random lick, anticipatory
lick, and water lick) between light and ultrasound stimulation. No significant difference

Figure 3-12. Water-licking behavior tests show that rats have similar behavioral responses to
light and ultrasound retinal stimulations. (a) Licking rates change along the time in each trial:
before, during and after the stimulation. (b) Anticipatory licking rate increased along the trial
number. (c) Raster plot of licking behavior during each trial. (d) The summary and statistical
analysis of licking rates in three stages.

52

was observed. The results mean that ultrasound retinal stimulation can evoke similar visual
signals as a pointed light and lead to the same behavioral responses. The results also rule
out the possibility that ultrasound-induced sounds [85] cause artificial neuron activities
because the subjects are only trained to anticipate water when there are visual signals. In
addition, since the normal rats were never trained by ultrasound stimulation, their water-
licking behavior cannot be a response to the mechanical pricking from ARF on the retina.
Water-licking experiment was also performed on RCS blind rats. However, since it is
impossible to use light stimulation to train blind subjects, ultrasound stimulation was used
in both the training and testing phases. Results are shown in the black lines of Fig. 3-13.
Fig. 3-13(a) shows that blind rats have increased licking rates with more trials, which
means rats can also learn the relationship between ultrasound stimulation and water drops.
Fig. 3-13(b) shows the average licking rate in each trial. Increased anticipatory licking rate
indicates that subjects are expecting water drops after receiving stimulations.
3.4 Discussion and Conclusion

Figure 3-13. Water-licking behavioral tests of blind rats. (a) Anticipatory licking rates in the
eight days training. The licking rates increase with more learning. The black line shows the
learning curve of normal rats, and the blue line is the learning curve of normal rats. (b) Licking
action happens before, during, and after the stimulation. Blind and normal rats show the similar
curves, indicating the same learning behavior.

53

This work is a follow-up of the work in Chapter 2 to further explore the feasibility and
efficacy of non-invasive ultrasound retinal prosthesis. The ultrasonic 2D array is necessary
for a practical prosthesis because it has the ability to generate arbitrary patterns and
dynamically change patterns at a high frame rate. Also, a 2D array provides the ability to
image the eyeball. Although the imaging quality is not as good as an imaging transducer,
it can still be utilized to locate the focal depth and calibrate the stimulation patterns. In this
work, we demonstrated that different patterns could be generated to stimulate the retina,
and the corresponding patterns were mapped from visual circuits (SC), indicating a
successful vision restoration was achieved. Also, ultrasound imaging is demonstrated to
locate the retina, guiding the stimulation focus.
We also investigate the frequency-dependent efficiency of ultrasound retinal stimulation.
The results are consistent with the previous ex-vivo study that higher frequency is more
efficient for retina stimulation, providing in-vivo evidence to the conclusion that ARF is
the physical mechanism of ultrasound retinal stimulation.
Safety concerns are also further investigated in this study. Long-term stimulation of two
weeks was conducted. By using multiple comprehensive investigation methods, including
Fundus imaging, OCT imaging, ERG measurements, and histology (H&E, GFAP, and
CD68 staining), the safety concerns are well addressed. No damage in morphology or
function was observed.
Finally, we performed the water-licking test on both normal and blind rats to investigate
the behavior response to ultrasound retinal stimulation. The positive results further support
the efficacy and validity of ultrasound retinal prostheses.
Although the ultrasound retinal prosthesis has been well investigated and demonstrated
in rodents, the morphological differences between rodent eyes and human eyes have to be
considered in conceiving a real ultrasonic retinal prosthesis for humans. These differences
provide both challenges and opportunities. For example, the human eye has a much larger
size than the rodent eye, so the ultrasound waves have to propagate a longer distance to
focus or form patterns at the retina. Diffraction effect and attenuation would be at a
different level and comprise the pattern quality. Meanwhile, the larger eyeball can wear a
larger ultrasound transducer size, meaning a larger aperture size and more transducer
elements, which are advantageous for the final pattern quality. On the other hand, human

54

eyes have macular (diameter ~5.5 mm) and fovea (diameter ~1.5 mm), which are small
portions of the retina but are responsible for the most important high-resolution central
vision. This means an ultrasonic prosthesis for humans only needs to focus on the fovea
instead of the whole retina, while it also requires a much higher spatial resolution.
There are also some other engineering challenges to getting a practical product of
ultrasonic retinal prosthesis. For example, a high-frequency (20-40 MHz) 2D array for
power generation has never been available. Also, to make the array wearable, the
supplementary minimized electric circuits, power supply, and heat dissipation of the final
devices are all challenging and not mature for now. However, there is no unsolvable
obstacle to these engineering challenges.

55

Chapter 4 Transcranial Focused Ultrasound for Non-invasive
Neuromodulation of the Visual Cortex
4.1 Introduction
Recent studies have shown that tFUS is a promising non-invasive technology for
cortical neuromodulation. Ultrasound waves pass through the skull and stimulate almost
any area of the brain with precise spatiotemporal resolution. Meanwhile, the safety of tFUS
has been proved in many in-vivo studies in various animal models [84, 98-104] and humans
[43, 46, 105-107]. While electrical stimulation methods offer high targeting specificity
and resolution, invasive cranial surgery is required. Further, electrode implantation and its
maintenance cause problems, especially for long-term deep brain stimulation [108, 109].
Other non-invasive brain stimulation methods such as temporally interfering electric fields
[36], repetitive transcranial magnetic stimulation (TMS), and transcranial direct current
stimulation (tDCS) [37, 38] have a relatively poor spatial resolution (on the order of
centimeters) compared to tFUS (on the order of few millimeters). Additionally, it is
difficult for TMS and tDCS to modulate deep subcortical areas.
Age-related macular degeneration (AMD) and retinitis pigmentosa (RP) are two
common outer retinal degenerative (RD) diseases that can produce severe vision loss. It is
estimated that nearly 30% of the U.S. population greater than 75 years of age have AMD,
10% of whom may become legally blind [22, 110, 111]. Currently, there is no cure for
blindness resulting from end-stage AMD and RP. Retinal prostheses use electrical
stimulation to directly elicit neural activity at the inner retina and promise the best short-
term strategy to provide partial restoration of sight to the blind [22, 25, 27, 112]. In retinal
prosthesis, electrically induced neural activities are transmitted through the optic nerve to
the higher visual areas of the brain to obtain visual perception. There are also visual
prosthetic devices based on directly stimulating the visual cortex, such as the Orion System
by Second Sights Inc. This cortical visual prosthesis converts images captured by a
miniature video camera mounted on a patient’s glasses into electrical pulses transmitted
wirelessly to an array of electrodes on the surface of the visual cortex. Although several
types of visual prostheses have been developed, they all are reported to have severe
limitations. Firstly, current technologies have limited spatial resolution due to the limited

56

number of stimulating electrodes. Secondly, invasive devices require complex and difficult
surgical implantation procedures. They also cause significant issues, including
encapsulation, electrode degradation, and interference with residual vision resulting from
limitations in biocompatibility and power supply. Blindness can also occur due to
irreversible and permanent inner retinal damages caused by glaucoma, optic neuropathy
diseases, or accidents. Treating the above conditions by stimulating the retina may not be
very effective. Direct implantation of the prosthesis into the VC is an alternate option that
has the advantage of avoiding device implantation in the delicate retinal tissue. However,
the majority of the issues associated with eye implantation could persist when implanting
a prosthesis in the brain. In addition, electrical stimulation in the brain may produce side
effects because of the strong current used in prostheses. Therefore, there is an unmet
clinical need for developing new techniques to cure blindness.
Non-invasive US visual prosthesis has great potential for restoring lost vision and is a
promising treatment for patients suffering from blindness. Some studies have shown that
tFUS can modulate the neuronal activities of the VC. Seung-Schik Yoo’s group used 5%
duty cycle tFUS to suppress visually evoked potentials (VEPs) in rats and elevated the
VEPs using higher duty cycles and stronger ultrasound intensity [113]. Human studies
conducted by the above group suggested that tFUS can stimulate the human VC, resulting
in the perception of phosphene and associated evoked potentials. The study also showed a
network of activated brain regions that are typically involved in visual and higher-order
cognitive processes [43]. In all the above studies, only normal animals and humans without
any visual disability were tested. So far, no studies have been conducted to restore vision
using tFUS by targeting the VC. In addition to this, the mechanism behind ultrasound
neuromodulation is still undetermined, making it difficult to optimize ultrasonic
parameters in various applications.
In this study, we use tFUS to stimulate the VC of normal rats and RD rats with severe
retinal degeneration that is considered to be blind. VC-evoked potentials were measured to
provide an electrophysiological assessment of brain activities. Different ultrasonic
waveforms were tested, and different types of responses were analyzed. Light-evoked
potentials were also measured for comparison with tFUS-evoked activities. Our results
suggested that tFUS stimulation of the VC in rats can evoke neuronal activities in both

57

normal and RD blind rats. Also, the blind rats showed significantly stronger responses to
ultrasound stimulation compared to the normal rats. The results also support the idea that
the predominant physical mechanism of ultrasound neuromodulation is the acoustic
radiation force (ARF).
4.2 Materials and Methods
4.2.1 Ultrasound Transducer and Waveform
A self-designed 0.5 MHz transducer was used to stimulate the VC, with a center
frequency of 0.5 MHz, 23 mm focal length, and an f-number of 0.7 for a minimized focal
area. A 3D-printed collimated cone was attached to the surface of the transducer for better
collimation and easier manipulation in experiments. The transducer was manipulated by a
5-axis stage to aim at VC. A function generator (AFG3252C, Tektronix, Beaverton, OR,
USA) was connected to a radiofrequency power amplifier (100A250A, Amplifier Research,
Souderton, PA, USA) to drive the transducer. The diagram of the experimental system is
shown in Fig. 4-1. Different types of waveforms were used in the three experimental groups:
1) A 15-ms-long continuous wave (CW) following a rest of 6 seconds.
2) Two 2-ms-long continuous waves following a rest of 6 seconds. The interval between
the two waves was 20 ms.
Figure 4-1. (a) Schematic diagram of the experimental system. (b) The surface topography of
the cortical areas of the left hemisphere of the rat. Each background grid has a length of 2 mm.
The primary visual cortex, area 17, is shown stippled. The exact position of the electrode is
shown by the red point. Ultrasound focal area is shown by the orange circle. (c) Top: Simulation
results of spatial distribution of acoustic intensity. Bottom: Simulation results of the ultrasound-
induced temperature increasing and its spatial distribution.

58

3) Pulsed waves (PW) with a high pulse repetition frequency (PRF) were used in group 3.
Five sets of parameters were used: 500 Hz PRF with 1-ms pulse, 500 Hz with 0.5-ms pulse,
333.3 Hz PRF with 1-ms pulse, 200 Hz PRF with 2-ms pulse, and 100 Hz PRF with 5-ms
pulse. All stimulations were conducted for 30 ms following a rest of 6 seconds.
The ultrasound field was measured by the hydrophone (HGL-0400, ONDA, Sunnyvale,
CA, USA) in a large water tank (free field). The ultrasonic spatial peak pulse average
intensity ISPPA used in all experiments was the same. The spatial-peak temporal-average
intensity (ISPTA) was defined as ISPTA = ISPPA * DC. DC is the duty cycle (%) of the
pulsed waveforms, which were different in the three groups.
To better illustrate the acoustic field and the ultrasound-induced temperature increase,
a finite-element simulation was conducted using finite-element analysis software.
(COMSOL 5.3a, COMSOL Inc., Burlington, MA, USA.)
4.2.2 Animal Preparation
All animal procedures were approved by the University of Southern California
Institutional Animal Care and Use Committee (IACUC). Seventeen rats were studied. Six
of them were normally sighted Long-Evan (LE) rats, and eleven were RD Royal College
of Surgeon (RCS) blind rats. Three normal rats and three blind rats were used in
experimental groups 1 & 2. Five blind rats were used in group 3. The RCS rats are
characterized by retinal pigment epithelium (RPE) dysfunction due to the deletion of the
Mer tyrosine kinase (MerTK) receptor that abolishes the internalization of photoreceptor
(PR) outer segments by RPE cells. All rats were male and about six-month-old. The rats
were anesthetized initially with an intraperitoneal injection of Ketamine/Xylazine (50-90
mg/kg, 5-10mg/kg), then with sevoflurane inhalation through a nose cone. [94] The eyes
were dilated using 1% tropicamide and 2.5% phenylephrine drops. The cranium was
exposed by removing the skin above the skull. A small cranial hole was made using a dental
drill. All procedures and experiments were performed in a dark room illuminated with dim
red light to minimize possible stimulation of the VC due to photoreceptor activation. The
space between the brain surface and transducer was filled using US gel.

4.2.3 Recording of VC-Evoked Potential Activities

59

A tungsten needle electrode (E363T, P1 Technologies, Roanoke, VA, USA) was
advanced into the visual cortex using the stereotactic apparatus. The electrode was aligned
to the ultrasound focal area as accurately as possible, as shown in Fig. 4-1(b). The reference
electrode was attached to the scalp, and the ground electrode was placed on the hindlimb.
Signals were recorded by a PowerLab data acquisition (DAQ) system (ADInstruments,
Sydney, Australia). The sampling frequency was 100 kHz. DAQ was synchronized with
ultrasound stimulation and light stimulation. To record light stimulation activities in the
brain, a full-field strobe flash using a Grass Photic stimulator (Grass Instrument Co., W.
Warwick, RI, USA) was delivered to the contralateral eye with a 6-second interstimulus
interval. The optical stimulation has a duration shorter than <1 ms. In all experiments,
recordings were repeated eight times. For each recording, the signals were averaged 64
times. Two filters, together with a 60Hz notch filter, were applied to all signals. The 300
Hz-25000 Hz filter highlights the short-time/high-frequency responses, which were
observed in only groups 1 & 2 (CW ultrasound). In contrast, the 0.1 Hz-300 Hz filter
highlights the long-time/low-frequency responses, which were observed only in group 3
(PW ultrasound). Light stimulation signals were filtered with a 60 Hz notch filter. The
amplitude of the evoked potentials was used to quantify the stimulation effects in this study.
Signal processing and statistical analysis were conducted using MATLAB 2017a
(Mathworks, Natick, MA, USA).
4.2.4 Experimental Design
The major difference between the three experimental groups was that different
ultrasound waveforms were used. All animals went through the same experimental
procedures: baseline recording, light stimulation, ultrasound stimulation, and no
stimulation control recording. There were 5 minutes intervals between each step. During
the control recording, the ultrasound was still on, but the transducer was oriented to a
completely different location (away from the brain).

4.3 Results
4.3.1 Ultrasound Field

60

The measured focused ultrasound pressure in the free field had a peak-to-peak amplitude
of 3.74 MPa, the intensity ISPPA was 115.8 W/cm
2
, and the mechanical index (MI) was 2.6.
The -6dB focal area had a lateral width of 2.4 mm, and an axial length of 5.1 mm. A
stimulated ultrasound field has been shown in Fig. 4-1(c). Ultrasound heating effect can be
neglected given the relatively low center frequency and short stimulation time (in the order
of milliseconds). In the simulation, the absorption coefficient was set to 5 Np/m, which is
the high limit of 0.5-MHz ultrasound. [114] The stimulation time was set to 100 ms, which
is much longer than the stimulation time in this study (<15 ms/6 s). In the bottom of Fig.
4-1(c), the color represents the increased temperature (K) caused by ultrasound stimulation.
With these loose conditions, the increased temperature is lower than 0.3℃.
The center frequency of ultrasound was chosen as 0.5 MHz mainly based on two aspects:
skull penetration and focal size. Lower frequency provides better penetration and worse
resolution. Considering the surface topography of the rat brain, as shown in Fig. 4-1(b), the
focus should have a diameter smaller than 4mm for a region-specific stimulation of the
Figure 4-2. Representative results evoked by continuous ultrasound stimulation from one normal
rat (a-d) and one blind rat (e-h). Gray lines show eight-times records and red lines are the averaged
records. Light stimulation was used to test the rats’ visual responsiveness. Normal rats responded
to the light (a) while blind rats failed to show any light responses (e). VC potential baseline
recorded from during control experiments (when transducer was focused to a different direction)
showed no responses in both rat groups (b)&(f). (c)&(g) show the responses with 15ms
ultrasound stimulation (shown by the red line) from the group 1. (d)&(h) show the responses from
the group 2.

61

visual cortex [115]. Higher-frequency ultrasound will be strongly distorted by the skull.
Therefore, 0.5 MHz is chosen as the center frequency in this study, which has also been
widely used in various ultrasound neuromodulation studies. For safety concerns, the
amplitude of the ultrasound was determined to be as low as possible to see clear responses.
Amplitude was the same in all experimental groups.
4.3.2 VC Responses
Representative results from one LE normal rat and one RCS blind rat are shown in Fig.
4-2. During US stimulation, 20-ms-long signals were recorded in group 1 rats, and 55-ms-
long signals were recorded in group 2 rats. Different durations of CW were tried in the
study. The durations from 2 ms to 15 ms caused similar responses, except that both ‘ON’
and ‘OFF’ responses can be observed in 15-ms-CW stimulation on blind rats and cannot
be observed in 2-ms-CW stimulation. Initially, light stimulation was used to test the rats’
visual sense, Figs. 4-2(a)&(e) for which 20-ms-long signals were recorded. The light
stimulation responses from the normal rat had a peak-to-peak amplitude of ~15 µV, while
the blind rats did not show any light-evoked cortical activities. US-evoked cortical

Figure 4-3. Comparison of response amplitude in groups 1&2 showing blind rats have
significantly stronger VC responses to US stimulation (p<0.001, two-sample t-test). In
experimental group 1, normal rats showed a response amplitude of 1.88±0.04 µV, whereas
blind rats had 4.87±0.38 µV. In group 2, the response amplitudes were 1.69±0.18 µV for
normal rats and 3.84±0.14 µV for blind RCS rats.

62

responses were obtained from both normal rats and blind rats. The peak-to-peak amplitude
of US responses is summarized in Fig. 4-3. The US responses from blind rats were
significantly stronger than those from normal LE rats (p<0.001, two-sample t-test).
Looking into the response onset latency and response duration, light responses had a
latency of ~1 ms and a duration of ~2 ms. In both LE and RCS rats, the US-evoked
activities showed latencies comparable to the light responses (~1 ms). The response
duration was considerably longer during US stimulation (~5 ms).
In group 3, 30-ms-long pulsed waves were used to stimulate the VC. Different sets of
parameters (duty cycle, duration, and PRF) were tested. Each set of parameters was tested
on one blind rat. US responses from different waveforms are shown in Fig. 4-4. 100-ms-

Figure 4-4. VC low-frequency responses to pulsed US stimulation with high PRF. US
stimulation duration was 30 ms, which shown by the red bar at the bottom. All signals were
averaged 512 times. Signals were recorded with different stimulation parameters: (a) 1-ms pulse
in every 2 ms (500 Hz PRF and 50% DC); (b) 0.5-ms pulse in every 2 ms (500 Hz PRF and 25%
DC); (c) 1-ms pulse in every 3 ms (333.3 Hz PRF and 33.3% DC); (d) 2-ms pulse in every 5 ms
(200 Hz PRF and 40% DC); (e) 5-ms pulse in every 10 ms (100 Hz PRF and 50% DC). Most
responses (a)-(d) had a similar waveform which was labeled by P1, P2, N1, N2 in (a).

63

long signals were recorded and averaged 512 times. Different duty cycles (DC), PRFs, and
pulse lengths were compared. The evoked potentials had a similar specific waveform:
starting with a weak ~10-ms negative peak (N1) followed by a strong ~10-ms positive peak
(P1), then a weaker negative peak (N2) and ending with a ~20-ms positive peak (P2). This
waveform pattern was observed in all cases except when a 5-ms pulse in every 10 ms (100
Hz PRF and 50% DC) was used (see Fig. 4-4e). Although the durations of the peaks were
similar, the amplitude of the peaks was different when different stimulation parameters
were used.
4.4 Discussion and Conclusion
Our study demonstrated that tFUS could evoke VC neuronal activities that are
comparable to light-stimulated responses. Although US-evoked VC potentials observed in
our study were weaker than light-stimulated responses, it is reasonable to predict that the
responses would be stronger if a stronger US intensity was used [53]. The results from
group 2 demonstrated that ultrasound VC stimulation could produce temporal resolution
shorter than 20 ms. Temporal resolution determines the available frame rate and is an
important parameter for developing successful visual prosthesis technology. As shown in
Fig. 4-2, the durations were always around 5 ms which provides a potential frame rate of
200 Hz. It should be noted that the responses in group 2 were slightly longer than the
responses in group 1, especially in blind rats. Presumably, this is a result of the overlap
between the US-evoked ‘ON’ and ‘OFF’ responses.
A key point of our work is the finding that US-evoked VC activities in blind RCS rats
were different from those of the normal LE rats. Under the same stimulus conditions, the
responses in blind rats were significantly stronger than that of normal rats. During US
stimulation, both ‘ON’ and ‘OFF’ responses were visible in blind rats, whereas only a
single peak (apparently ‘on’ response) was recorded from normal rats. One possible
explanation for the differences is that the stimulation sensitivity of blind rats’ VC has been
changed due to prolonged visual deprivation. It is highly unlikely that such differences can
be attributed to the difference in rat strains. Further investigations using the same animal
strains can provide a better conclusion in this regard.

64

Another major finding of this study is that different ultrasound waveforms can evoke
different types of VC responses. In experimental group 3, PW was used to stimulate the
VC. Besides the ~5 ms duration responses, which can be shown using a high-pass filter, a
different type of long-duration low-frequency response was acquired. As shown in Fig. 4,
this type of response had a duration longer than 20ms with four peaks (N1, P1, N2, P2) at
specific time points. The above waveform pattern persisted at different stimulus parameters
until the PRF went down to 100 Hz (Fig. 4-4e). Different DCs were tested at 100 Hz PRF,
but the typical waveform was never observed. PRF down to 50 Hz was also tested to
confirm that the above waveform cannot be observed at lower PRF. Although the data
obtained from group 3 was not sufficient to show statistically significant differences based
on various parameters tested (pulse length, PRF, DC), it showed a clear trend to support
the hypothesis that PRF larger than 100 Hz is essential to obtain the low-frequency
responses. In groups 1 & 2, 2-ms waves and 15-ms waves were used to show similar short-
duration responses. Kim Butt Pauly’s group used 80-ms-long US stimulation to the
auditory cortex to obtain ‘US on & off’ short-duration responses that are comparable to our
observation [116]. It can be inferred that changes in the duration of CW stimulation (at
least within the range of 100 ms) may not cause changes in the response pattern like PW.
In this study, the acoustic field was measured in a free field, and the MI was 2.6, which
is a little higher than the U.S. Food and Drug Administration (FDA)’s limitation of
ultrasound imaging 1.9. However, since the MI was measured in a free field, it is reasonable
to predict that the real acoustic energy that reached the brain would be weaker than the
reported pressure due to the distortion caused by skull and tissue absorption. It is estimated
that the ultrasound transmission factor through the adult rat skull is around 0.5-0.7 at 0.5
MHz [117, 118], which will reduce the MI from 2.6 to 1.3-1.8. (<1.9). Therefore,
considering the distorting caused by the skull, the MI in this study should be within the
FDA limitation.
There are some limitations in this study that need to be addressed in future investigations.
Rats of different sex and age should be compared. Additional sets of stimulus parameters
are required for PW stimulation to validate the effect of individual stimulus conditions. We
are not ruling out the possibility of backward projections along visual pathways. This can
be investigated by suitable staining techniques like C-FOS detection to map the activated

65

neurons. It may also help to understand the extent of spatial resolution. We plan to
implement these for our future experiments.
There have been arguments about whether the ultrasound neuromodulation effect in the
VC may be partially due to the auditory response to the audible sound caused by pulsed
ultrasound or mode conversion [85]. The study from Hubert Lim’s group suggested that
the removal of cochlear fluids (CF) or transection of the auditory nerves would eliminate
the US neuromodulation effect [84]. But Kim’s research disproved this argument by
eliminating peripheral auditory pathway activation and obtaining motor responses during
US stimulation [116]. Although genetically deaf rats or CF removal is required in our future
studies to completely eliminate the auditory pathway effects, we conducted some
preliminary experiments to rule out its contribution to VC activation. We stimulated the
auditory cortex but did not observe any responses at VC. One of the potential factors that
causes the arguments and inconsistency among different studies is the stimulation area
relative to the animal size, which is related to the animal model, ultrasound frequency, and
ultrasound focusing. Minimized ultrasound focal points and localized stimulation areas are
desirable to minimize the experimental artifacts.
The physical mechanism of the in-vivo ultrasound neuromodulation effect is another
significant but undetermined question. Understanding the mechanisms of ultrasound
neuromodulation can help optimize ultrasonic parameters in various applications and
scientific research. Recent studies using excised retinas support the idea that acoustic
radiation force (ARF) is the main mechanism for ultrasound-evoked neuronal activities [53]
ex vivo. The above study ruled out the cavitation effect by comparing the stimulation effect
induced by different-frequency ultrasounds with the same intensity. However, the study
did not discuss the acoustic oscillation effect, and the mechanism could be different under
in-vivo conditions. Our in vivo results suggest that ARF is the predominant mechanism for
ultrasound neuromodulation from another perspective. The two phenomena observed in
our experiments - the presence of ‘US on & off’ responses and response changes caused
by changes in the PW and CW, support the ARF hypothesis. The candidates for the
physical mechanism of US stimulation are cavitation, acoustic oscillation, thermal effect,
and ARF. However, the thermal effect is a gradual and accumulative effect depositing
energy in brain tissues, with which PW and CW should not cause different phenomena

66

[119]. Also, CW should have a stronger thermal effect than PW, which was not observed
in our experiments. The cavitation effect is the interaction between ultrasound waves and
bubbles, where bubbles can be generated if the acoustic pressure has sufficiently negative
peaks. The size of the bubble oscillates as the localized pressure changes sinusoidally.
Transient cavitation happens when the size expansion is at least double, at which point the
bubbles collapse violently, causing a destructive event [52]. In stable cavitation, the size
change is smaller, and the bubbles do not burst, which is hypothesized to produce stable
neuromodulation [53]. Similarly, acoustic oscillation is the intrinsic linear mechanical
effect of acoustic waves, which causes tissues sinusoidal oscillations with a much weaker
amplitude compared to the cavitation effect. ARF is a nonlinear acoustic effect that
generates a non-oscillating force and causes a unidirectional displacement in biological
tissues [120]. All mechanical acoustic effects are hypothesized to affect local neural ion
channels, facilitate cellular ion flux and cause electrical activities of neurons [62, 121-123].
The key difference that highlights the ARF is the fact that ARF has a non-oscillating
unidirectional effect, while stable cavitation and acoustic oscillation are sinusoidal and
periodic effects. If stable cavitation or acoustic oscillation is the mechanism, the ion influx
and outflux should be periodic as well. It is less likely for periodic effects to cause
responses only during US ‘ON’ and ‘OFF’, and no responses in between. If ARF is the
mechanism of neuronal stimulation, the occurrence of the above phenomena is more
reasonable. Because the US on can push the neuron or ion channels and cause the
measurable electrical potentials, the local potentials will regain the balance (absence of
response) until the US is off, which can well explain the ‘on’/‘off’ phenomena observed in
our study. We caution that the statement about ARF as the principal physical mechanism
of neuromodulation is only a result-oriented interference rather than a well-reasoned
conclusion from elaborated experiments. However, we believe that results from this study
deserve to be considered in further discussions and future studies on the mechanisms of
ultrasound neuromodulation.
Although the size of the ultrasound focus in this study is not fine enough for a practical
visual prosthesis, ultrasound has the potential to resolve micrometer-level by increasing
center frequency and enhancing the transducer’s focusing performance. In addition, MRI-

67

guided/CT-guided ultrasound technology can enhance the focus region and resolution in
further studies.
The feasibility of tFUS-based cortical visual prosthesis is studied and discussed here.
tFUS can be used to evoke neural activities in the VC of normal and blind rats. Different
types of VC responses could be evoked by different US stimulus waveforms. ARF is
inferred to be the predominant physical mechanism of ultrasound neuromodulation. Based
on our study, ultrasonic neuromodulation is a promising technology for the development
of non-invasive cortical visual prostheses. It can be applied to treat blindness caused by
irreversible inner retinal damages (optic neuropathy or accidents) where retinal prosthesis
cannot be effective.

68

Chapter 5 Summary and Future Work
5.1 Summary
Vision is one of the most essential sensors of humans. However, there is still no cure
for several types of blindness. Electrode-based prostheses are the best near-term strategy.
However, intrinsic invasiveness highly limits its performance and patient acceptance. This
research focuses on the exploration and development of noninvasive ultrasonic visual
prostheses. In Chapter 2, using normal-sighted and retinal degenerative blind rats, the
ability of US stimulation was successfully demonstrated to evoke neural activity in the
retina. Using MEA recordings at SC, the spatiotemporal resolution of US retinal
stimulation was quantified, showing the promising application of US retinal stimulation as
a novel and noninvasive retinal prosthesis.
After demonstrating that US retinal stimulation can activate neurons in blind rats, the
feasibility of a practical ultrasonic retinal prosthesis was further explored. In Chapter 3, a
16 × 16 4.5-MHz ultrasonic 2D array was designed and fabricated for vision restoration. It
was successfully shown that the 2D array could dynamically generate arbitrary patterns
and stimulates the retina as well as evoke corresponding visual pattern signals in the visual
circuit in the brain. Furthermore, the 2D array can provide a sufficient guiding image to
locate the retina and guide pattern focus before the stimulation. The safety of ultrasonic
retinal prostheses was further investigated using comprehensive imaging and histology
methods. Positive results confirmed that the US power is within a safe range. Finally,
animal behavior tests were conducted to show direct and high-level behavior validation of
US retinal prosthesis.
While retinal prosthesis can use natural visual signal processing in visual circuits, some
blind patients with damaged ONs cannot benefit from retinal stimulation. In Chapter 4,
transcranial US stimulation on the VC was further applied, and it was demonstrated that
localized neurons could be activated. The safety of US stimulation was confirmed. By
measuring the temperature increase and US power, our in vivo results also indicated that
the ARF is the primary physical mechanism for US stimulation.

5.2 Future Work

69

The work shown in this dissertation is a proof of concept of US stimulation as visual
prostheses and illustrates its favorable prospects for clinical applications. The future
development and investigation of US stimulation will be based on three aspects.
5.2.1 Wearable High-frequency Ultrasound Array-based Retinal Prosthesis
Although we have conducted a comprehensive investigation of the feasibility of US
retinal prosthesis and proved its efficacy, the design and development of a wearable device
is still difficult. Predictable obstacles include thickness, power consumption, heating
dissipation, attachment, and US coupling to the cornea. Many engineering efforts must be
devoted to developing and optimizing the final prosthesis device.
5.2.2 High-frequency US VC Stimulation
Another aspect is to continue working on US stimulation on the VC. The previous work
used a 0.5-MHz transducer. The resolution is too poor for a possible visual prosthesis. In
addition, the recording was conducted using a low-impedance single electrode. No neural
spikes were recorded. In the next step, higher-frequency US (ranging from 2 to 8 MHz)
with MEA recording will be used to stimulate and record the neuron at the VC.
5.2.3 Study of Biophysical Mechanism
Most of these studies were focused on the feasibility and applications of US stimulation.
Given the favorable results, the underlying biophysical mechanism of US stimulation will
be investigated in the next step. Isolated ganglion cells, extracted retina, or the sliced VC
from mice with GCaMP6 expression will be stimulated, and calcium imaging will be
employed to monitor neural activities. By blocking various mechanosensitive ion channels,
the roles of mechanosensitive ion channels in US stimulation will be investigated.

70

BIBLIOGRAPHY
[1] C. N. Shealy, J. T. Mortimer, and J. B. Reswick, “Electrical inhibition of pain by
stimulation of the dorsal columns: preliminary clinical report,” Anesthesia & Analgesia,
vol. 46, no. 4, pp. 489-491, 1967.
[2] N. A. Mekhail, J. Cheng, S. Narouze, L. Kapural, M. N. Mekhail, and T. Deer, “Clinical
applications of neurostimulation: forty years later,” Pain Practice, vol. 10, no. 2, pp. 103-
112, 2010.
[3] M. I. Hariz, P. Blomstedt, and L. Zrinzo, “Deep brain stimulation between 1947 and
1987: the untold story,” Neurosurgical focus, vol. 29, no. 2, pp. E1, 2010.
[4] A. L. Benabid, “Deep brain stimulation for Parkinson’s disease,” Current opinion in
neurobiology, vol. 13, no. 6, pp. 696-706, 2003.
[5] P. Limousin, P. Krack, P. Pollak, A. Benazzouz, C. Ardouin, D. Hoffmann, and A.-L.
Benabid, “Electrical stimulation of the subthalamic nucleus in advanced Parkinson's
disease,” New England Journal of Medicine, vol. 339, no. 16, pp. 1105-1111, 1998.
[6] W. H. Theodore, and R. S. Fisher, “Brain stimulation for epilepsy,” The Lancet
Neurology, vol. 3, no. 2, pp. 111-118, 2004.
[7] A. Schulze-Bonhage, “Brain stimulation as a neuromodulatory epilepsy therapy,”
Seizure, vol. 44, pp. 169-175, 2017.
[8] H. S. Mayberg, A. M. Lozano, V. Voon, H. E. McNeely, D. Seminowicz, C. Hamani, J.
M. Schwalb, and S. H. Kennedy, “Deep brain stimulation for treatment-resistant
depression,” Neuron, vol. 45, no. 5, pp. 651-660, 2005.
[9] B. D. Greenberg, D. A. Malone, G. M. Friehs, A. R. Rezai, C. S. Kubu, P. F. Malloy, S.
P. Salloway, M. S. Okun, W. K. Goodman, and S. A. Rasmussen, “Three-year outcomes
in deep brain stimulation for highly resistant obsessive–compulsive disorder,”
Neuropsychopharmacology, vol. 31, no. 11, pp. 2384-2393, 2006.
[10] M. Figee, J. Luigjes, R. Smolders, C.-E. Valencia-Alfonso, G. Van Wingen, B. De
Kwaasteniet, M. Mantione, P. Ooms, P. De Koning, and N. Vulink, “Deep brain
stimulation restores frontostriatal network activity in obsessive-compulsive disorder,”
Nature neuroscience, vol. 16, no. 4, pp. 386-387, 2013.
[11] C. H. Halpern, J. A. Wolf, T. L. Bale, A. J. Stunkard, S. F. Danish, M. Grossman, J. L.
Jaggi, M. S. Grady, and G. H. Baltuch, “Deep brain stimulation in the treatment of
obesity: a review,” Journal of neurosurgery, vol. 109, no. 4, pp. 625-634, 2008.
[12] A. W. Laxton, D. F. Tang‐Wai, M. P. McAndrews, D. Zumsteg, R. Wennberg, R. Keren,
J. Wherrett, G. Naglie, C. Hamani, and G. S. Smith, “A phase I trial of deep brain
stimulation of memory circuits in Alzheimer's disease,” Annals of neurology, vol. 68, no.
4, pp. 521-534, 2010.

71

[13] A. W. Laxton, and A. M. Lozano, “Deep brain stimulation for the treatment of Alzheimer
disease and dementias,” World neurosurgery, vol. 80, no. 3-4, pp. S28. e1-S28. e8, 2013.
[14] J. N. Fayad, S. R. Otto, R. V. Shannon, and D. E. Brackmann, “Cochlear and Brainstem
Auditory Prostheses “Neural Interface for Hearing Restoration: Cochlear and Brain Stem
Implants”,” Proceedings of the IEEE, vol. 96, no. 7, pp. 1085-1095, 2008.
[15] A. R. Møller, “History of cochlear implants and auditory brainstem implants,” Cochlear
and brainstem implants, vol. 64, pp. 1-10, 2006.
[16] H. H. Lim, M. Lenarz, and T. Lenarz, “Auditory midbrain implant: a review,” Trends in
amplification, vol. 13, no. 3, pp. 149-180, 2009.
[17] T. Lenarz, H. H. Lim, G. Reuter, J. F. Patrick, and M. Lenarz, “The auditory midbrain
implant: a new auditory prosthesis for neural deafness-concept and device description,”
Otology & Neurotology, vol. 27, no. 6, pp. 838-843, 2006.
[18] M. S. Humayun, J. D. Dorn, L. Da Cruz, G. Dagnelie, J.-A. Sahel, P. E. Stanga, A. V.
Cideciyan, J. L. Duncan, D. Eliott, and E. Filley, “Interim results from the international
trial of Second Sight's visual prosthesis,” Ophthalmology, vol. 119, no. 4, pp. 779-788,
2012.
[19] J. D. Weiland, and M. S. Humayun, “Visual prosthesis,” Proceedings of the IEEE, vol.
96, no. 7, pp. 1076-1084, 2008.
[20] M. S. Beauchamp, D. Oswalt, P. Sun, B. L. Foster, J. F. Magnotti, S. Niketeghad, N.
Pouratian, W. H. Bosking, and D. Yoshor, “Dynamic stimulation of visual cortex
produces form vision in sighted and blind humans,” Cell, vol. 181, no. 4, pp. 774-783. e5,
2020.
[21] V. Gaillet, A. Cutrone, F. Artoni, P. Vagni, A. M. Pratiwi, S. A. Romero, D. L. Di Paola,
S. Micera, and D. Ghezzi, “Spatially selective activation of the visual cortex via
intraneural stimulation of the optic nerve,” Nature biomedical engineering, vol. 4, no. 2,
pp. 181-194, 2020.
[22] J. D. Weiland, W. Liu, and M. S. Humayun, “Retinal prosthesis,” Annu. Rev. Biomed.
Eng., vol. 7, pp. 361-401, 2005.
[23] R. A. Schmidt, U. Jonas, K. A. OLESON, R. A. JANKNEGT, M. M. HASSOUNA, S.
W. SIEGEL, and P. E. van KERREBROECK, “Sacral nerve stimulation for treatment of
refractory urinary urge incontinence,” The Journal of urology, vol. 162, no. 2, pp. 352-
357, 1999.
[24] N. Rijkhoff, H. Wijkstra, P. Van Kerrebroeck, and F. Debruyne, “Urinary bladder control
by electrical stimulation: review of electrical stimulation techniques in spinal cord
injury,” Neurourology and urodynamics, vol. 16, no. 1, pp. 39-53, 1997.

72

[25] J. O. Winter, S. F. Cogan, and J. F. Rizzo, “Retinal prostheses: current challenges and
future outlook,” Journal of Biomaterials Science, Polymer Edition, vol. 18, no. 8, pp.
1031-1055, 2007.
[26] L. da Cruz, J. D. Dorn, M. S. Humayun, G. Dagnelie, J. Handa, P.-O. Barale, J.-A. Sahel,
P. E. Stanga, F. Hafezi, and A. B. Safran, “Five-year safety and performance results from
the Argus II retinal prosthesis system clinical trial,” Ophthalmology, vol. 123, no. 10, pp.
2248-2254, 2016.
[27] J. Dowling, “Current and future prospects for optoelectronic retinal prostheses,” Eye, vol.
23, no. 10, pp. 1999-2005, 2009.
[28] K. Mathieson, J. Loudin, G. Goetz, P. Huie, L. Wang, T. I. Kamins, L. Galambos, R.
Smith, J. S. Harris, and A. Sher, “Photovoltaic retinal prosthesis with high pixel density,”
Nature photonics, vol. 6, no. 6, pp. 391-397, 2012.
[29] C. Veraart, C. Raftopoulos, J. T. Mortimer, J. Delbeke, D. Pins, G. Michaux, A.
Vanlierde, S. Parrini, and M.-C. Wanet-Defalque, “Visual sensations produced by optic
nerve stimulation using an implanted self-sizing spiral cuff electrode,” Brain research,
vol. 813, no. 1, pp. 181-186, 1998.
[30] J. Delbeke, M. Oozeer, and C. Veraart, “Position, size and luminosity of phosphenes
generated by direct optic nerve stimulation,” Vision research, vol. 43, no. 9, pp. 1091-
1102, 2003.
[31] V. Gaillet, A. Cutrone, F. Artoni, P. Vagni, A. Mega Pratiwi, S. A. Romero, D. Lipucci
Di Paola, S. Micera, and D. Ghezzi, “Spatially selective activation of the visual cortex via
intraneural stimulation of the optic nerve,” Nature biomedical engineering, vol. 4, no. 2,
pp. 181-194, 2020.
[32] P. M. Lewis, H. M. Ackland, A. J. Lowery, and J. V. Rosenfeld, “Restoration of vision in
blind individuals using bionic devices: a review with a focus on cortical visual
prostheses,” Brain research, vol. 1595, pp. 51-73, 2015.
[33] M. Bak, J. Girvin, F. Hambrecht, C. Kufta, G. Loeb, and E. Schmidt, “Visual sensations
produced by intracortical microstimulation of the human occipital cortex,” Medical and
Biological Engineering and Computing, vol. 28, no. 3, pp. 257-259, 1990.
[34] E. M. Schmidt, M. J. Bak, F. Hambrecht, C. V. Kufta, D. O'rourke, and P. Vallabhanath,
“Feasibility of a visual prosthesis for the blind based on intracortical micro stimulation of
the visual cortex,” Brain, vol. 119, no. 2, pp. 507-522, 1996.
[35] K. Cha, K. W. Horch, R. A. Normann, and D. K. Boman, “Reading speed with a
pixelized vision system,” JOSA A, vol. 9, no. 5, pp. 673-677, 1992.
[36] N. Grossman, D. Bono, N. Dedic, S. B. Kodandaramaiah, A. Rudenko, H.-J. Suk, A. M.
Cassara, E. Neufeld, N. Kuster, and L.-H. Tsai, “Noninvasive deep brain stimulation via
temporally interfering electric fields,” Cell, vol. 169, no. 6, pp. 1029-1041. e16, 2017.

73

[37] F. Fregni, and A. Pascual-Leone, “Technology insight: noninvasive brain stimulation in
neurology—perspectives on the therapeutic potential of rTMS and tDCS,” Nature
clinical practice Neurology, vol. 3, no. 7, pp. 383-393, 2007.
[38] M. S. George, and G. Aston-Jones, “Noninvasive techniques for probing neurocircuitry
and treating illness: vagus nerve stimulation (VNS), transcranial magnetic stimulation
(TMS) and transcranial direct current stimulation (tDCS),” Neuropsychopharmacology,
vol. 35, no. 1, pp. 301-316, 2010.
[39] K. Deisseroth, “Optogenetics,” Nature methods, vol. 8, no. 1, pp. 26-29, 2011.
[40] L. Fenno, O. Yizhar, and K. Deisseroth, “The development and application of
optogenetics,” Annual review of neuroscience, vol. 34, 2011.
[41] S. Ibsen, A. Tong, C. Schutt, S. Esener, and S. H. Chalasani, “Sonogenetics is a non-
invasive approach to activating neurons in Caenorhabditis elegans,” Nature
communications, vol. 6, no. 1, pp. 1-12, 2015.
[42] D. Maresca, A. Lakshmanan, M. Abedi, A. Bar-Zion, A. Farhadi, G. J. Lu, J. O.
Szablowski, D. Wu, S. Yoo, and M. G. Shapiro, “Biomolecular ultrasound and
sonogenetics,” Annual review of chemical and biomolecular engineering, vol. 9, pp. 229-
252, 2018.
[43] W. Lee, H.-C. Kim, Y. Jung, Y. A. Chung, I.-U. Song, J.-H. Lee, and S.-S. Yoo,
“Transcranial focused ultrasound stimulation of human primary visual cortex,” Scientific
reports, vol. 6, no. 1, pp. 1-12, 2016.
[44] J. Blackmore, S. Shrivastava, J. Sallet, C. R. Butler, and R. O. Cleveland, “Ultrasound
neuromodulation: a review of results, mechanisms and safety,” Ultrasound in medicine &
biology, vol. 45, no. 7, pp. 1509-1536, 2019.
[45] Y. Tufail, A. Yoshihiro, S. Pati, M. M. Li, and W. J. Tyler, “Ultrasonic neuromodulation
by brain stimulation with transcranial ultrasound,” nature protocols, vol. 6, no. 9, pp.
1453-1470, 2011.
[46] W. Legon, T. F. Sato, A. Opitz, J. Mueller, A. Barbour, A. Williams, and W. J. Tyler,
“Transcranial focused ultrasound modulates the activity of primary somatosensory cortex
in humans,” Nature neuroscience, vol. 17, no. 2, pp. 322-329, 2014.
[47] D. Nelidova, R. K. Morikawa, C. S. Cowan, Z. Raics, D. Goldblum, H. P. Scholl, T.
Szikra, A. Szabo, D. Hillier, and B. Roska, “Restoring light sensitivity using tunable
near-infrared sensors,” Science, vol. 368, no. 6495, pp. 1108-1113, 2020.
[48] D. Liao, M.-Y. Hsiao, G. Xiang, and P. Zhong, “Optimal pulse length of insonification
for Piezo1 activation and intracellular calcium response,” Scientific reports, vol. 11,
2021.

74

[49] Y. Yang, C. P. Pacia, D. Ye, L. Zhu, H. Baek, Y. Yue, J. Yuan, M. J. Miller, J. Cui, and
J. P. Culver, “Sonogenetics for noninvasive and cellular-level neuromodulation in rodent
brain,” bioRxiv, 2020.
[50] E. N. Harvey, “The effect of high frequency sound waves on heart muscle and other
irritable tissues,” American Journal of Physiology-Legacy Content, vol. 91, no. 1, pp.
284-290, 1929.
[51] F. Fry, H. Ades, and W. Fry, “Production of reversible changes in the central nervous
system by ultrasound,” Science, vol. 127, no. 3289, pp. 83-84, 1958.
[52] E. A. Neppiras, “Acoustic cavitation,” Physics reports, vol. 61, no. 3, pp. 159-251, 1980.
[53] M. D. Menz, P. Ye, K. Firouzi, A. Nikoozadeh, K. B. Pauly, P. Khuri-Yakub, and S. A.
Baccus, “Radiation force as a physical mechanism for ultrasonic neurostimulation of the
ex vivo retina,” Journal of Neuroscience, vol. 39, no. 32, pp. 6251-6264, 2019.
[54] M. L. Palmeri, A. C. Sharma, R. R. Bouchard, R. W. Nightingale, and K. R. Nightingale,
“A finite-element method model of soft tissue response to impulsive acoustic radiation
force,” IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52,
no. 10, pp. 1699-1712, 2005.
[55] G. Torr, “The acoustic radiation force,” American Journal of Physics, vol. 52, no. 5, pp.
402-408, 1984.
[56] T. GE, “Preliminary report on a retinal stimulator,” The British journal of physiological
optics, vol. 13, no. 2, pp. 102-105, 1956.
[57] A. K. Ahuja, J. Dorn, A. Caspi, M. McMahon, G. Dagnelie, P. Stanga, M. Humayun, R.
Greenberg, and A. I. S. Group, “Blind subjects implanted with the Argus II retinal
prosthesis are able to improve performance in a spatial-motor task,” British Journal of
Ophthalmology, vol. 95, no. 4, pp. 539-543, 2011.
[58] J. M. Barrett, R. Berlinguer-Palmini, and P. Degenaar, “Optogenetic approaches to retinal
prosthesis,” Visual neuroscience, vol. 31, no. 4-5, pp. 345-354, 2014.
[59] J.-A. Sahel, and B. Roska, “Gene therapy for blindness,” Annual review of neuroscience,
vol. 36, pp. 467-488, 2013.
[60] I. Tochitsky, M. A. Kienzler, E. Isacoff, and R. H. Kramer, “Restoring vision to the blind
with chemical photoswitches,” Chemical reviews, vol. 118, no. 21, pp. 10748-10773,
2018.
[61] Y. Lu, B. Brommer, X. Tian, A. Krishnan, M. Meer, C. Wang, D. L. Vera, Q. Zeng, D.
Yu, and M. S. Bonkowski, “Reprogramming to recover youthful epigenetic information
and restore vision,” Nature, vol. 588, no. 7836, pp. 124-129, 2020.

75

[62] A. Fomenko, C. Neudorfer, R. F. Dallapiazza, S. K. Kalia, and A. M. Lozano, “Low-
intensity ultrasound neuromodulation: An overview of mechanisms and emerging human
applications,” Brain stimulation, vol. 11, no. 6, pp. 1209-1217, 2018.
[63] H. A. S. Kamimura, A. Conti, N. Toschi, and E. Konofagou, “Ultrasound
neuromodulation: mechanisms and the potential of multi-modal stimulation for neuronal
function assessment,” Frontiers in physics, vol. 8, pp. 150, 2020.
[64] M. E. Downs, T. Teichert, A. Buch, M. E. Karakatsani, C. Sierra, S. Chen, E. E.
Konofagou, and V. P. Ferrera, “Toward a cognitive neural prosthesis using focused
ultrasound,” Frontiers in neuroscience, vol. 11, pp. 607, 2017.
[65] H. Jeong, J. J. Im, J.-S. Park, S.-H. Na, W. Lee, S.-S. Yoo, I.-U. Song, and Y.-A. Chung,
“A pilot clinical study of low-intensity transcranial focused ultrasound in Alzheimer's
disease,” Ultrasonography (Seoul, Korea), 2021.
[66] O. Naor, Y. Hertzberg, E. Zemel, E. Kimmel, and S. Shoham, “Towards multifocal
ultrasonic neural stimulation II: design considerations for an acoustic retinal prosthesis,”
Journal of neural engineering, vol. 9, no. 2, pp. 026006, 2012.
[67] Q. Jiang, G. Li, H. Zhao, W. Sheng, L. Yue, M. Su, S. Weng, L. L.-H. Chan, Q. Zhou,
and M. S. Humayun, “Temporal neuromodulation of retinal ganglion cells by low-
frequency focused ultrasound stimulation,” IEEE Transactions on Neural Systems and
Rehabilitation Engineering, vol. 26, no. 5, pp. 969-976, 2018.
[68] M. D. Menz, Ö. Oralkan, P. T. Khuri-Yakub, and S. A. Baccus, “Precise neural
stimulation in the retina using focused ultrasound,” Journal of Neuroscience, vol. 33, no.
10, pp. 4550-4560, 2013.
[69] D. C. Lozano, and M. D. Twa, “Development of a rat schematic eye from in vivo
biometry and the correction of lateral magnification in SD-OCT imaging,” Investigative
ophthalmology & visual science, vol. 54, no. 9, pp. 6446-6455, 2013.
[70] D. Schulz, M. E. Iliev, B. E. Frueh, and D. Goldblum, “In vivo pachymetry in normal
eyes of rats, mice and rabbits with the optical low coherence reflectometer,” Vision
research, vol. 43, no. 6, pp. 723-728, 2003.
[71] M. Nabili, C. Geist, and V. Zderic, “Thermal safety of ultrasound‐enhanced ocular drug
delivery: A modeling study,” Medical physics, vol. 42, no. 10, pp. 5604-5615, 2015.
[72] T. Şen, O. Tüfekçioğlu, and Y. Koza, “Mechanical index,” Anatolian journal of
cardiology, vol. 15, no. 4, pp. 334, 2015.
[73] B. B. Thomas, M. J. Seiler, S. R. Sadda, and R. B. Aramant, “Superior colliculus
responses to light–preserved by transplantation in a slow degeneration rat model,”
Experimental eye research, vol. 79, no. 1, pp. 29-39, 2004.
[74] H. Kanda, T. Morimoto, T. Fujikado, Y. Tano, Y. Fukuda, and H. Sawai,
“Electrophysiological studies of the feasibility of suprachoroidal-transretinal stimulation

76

for artificial vision in normal and RCS rats,” Investigative ophthalmology & visual
science, vol. 45, no. 2, pp. 560-566, 2004.
[75] P. Cesare, A. Moriondo, V. Vellani, and P. McNaughton, “Ion channels gated by heat,”
Proceedings of the National Academy of Sciences, vol. 96, no. 14, pp. 7658-7663, 1999.
[76] O.-J. Grüsser, U. Grüsser-Cornehls, R. Kusel, and A. Przybyszewski, “Responses of
retinal ganglion cells to eyeball deformation: a neurophysiological basis for “pressure
phosphenes”,” Vision research, vol. 29, no. 2, pp. 181-194, 1989.
[77] O.-J. Grüsser, M. Hagner, and A. Przybyszewski, “The effect of dark adaptation on the
responses of cat retinal ganglion cells to eyeball deformation,” Vision research, vol. 29,
no. 9, pp. 1059-1068, 1989.
[78] D. A. Ryskamp, P. Witkovsky, P. Barabas, W. Huang, C. Koehler, N. P. Akimov, S. H.
Lee, S. Chauhan, W. Xing, and R. C. Rentería, “The polymodal ion channel transient
receptor potential vanilloid 4 modulates calcium flux, spiking rate, and apoptosis of
mouse retinal ganglion cells,” Journal of Neuroscience, vol. 31, no. 19, pp. 7089-7101,
2011.
[79] C. M. Rountree, C. Meng, J. B. Troy, and L. Saggere, “Mechanical stimulation of the
retina: therapeutic feasibility and cellular mechanism,” IEEE Transactions on Neural
Systems and Rehabilitation Engineering, vol. 26, no. 5, pp. 1075-1083, 2018.
[80] S. Yoo, D. R. Mittelstein, R. C. Hurt, J. J. Lacroix, and M. G. Shapiro, “Focused
ultrasound excites neurons via mechanosensitive calcium accumulation and ion channel
amplification,” bioRxiv, 2020.
[81] B. Sorum, R. A. Rietmeijer, K. Gopakumar, H. Adesnik, and S. G. Brohawn, “Ultrasound
activates mechanosensitive TRAAK K+ channels through the lipid membrane,”
Proceedings of the National Academy of Sciences, vol. 118, no. 6, 2021.
[82] W.-H. Liao, M.-Y. Hsiao, Y. Kung, H.-L. Liu, J.-C. Béra, C. Inserra, and W.-S. Chen,
“TRPV4 promotes acoustic wave-mediated BBB opening via Ca2+/PKC-δ pathway,”
Journal of Advanced Research, vol. 26, pp. 15-28, 2020.
[83] J. Kubanek, P. Shukla, A. Das, S. A. Baccus, and M. B. Goodman, “Ultrasound elicits
behavioral responses through mechanical effects on neurons and ion channels in a simple
nervous system,” Journal of Neuroscience, vol. 38, no. 12, pp. 3081-3091, 2018.
[84] H. Guo, M. Hamilton II, S. J. Offutt, C. D. Gloeckner, T. Li, Y. Kim, W. Legon, J. K.
Alford, and H. H. Lim, “Ultrasound produces extensive brain activation via a cochlear
pathway,” Neuron, vol. 98, no. 5, pp. 1020-1030. e4, 2018.
[85] T. Sato, M. G. Shapiro, and D. Y. Tsao, “Ultrasonic neuromodulation causes widespread
cortical activation via an indirect auditory mechanism,” Neuron, vol. 98, no. 5, pp. 1031-
1041. e5, 2018.

77

[86] T. R. Magee, and A. H. Davies, “Auditory phenomena during transcranial Doppler
insonation of the basilar artery,” Journal of ultrasound in medicine, vol. 12, no. 12, pp.
747-750, 1993.
[87] K. Matsushima, and T. Shimobaba, “Band-limited angular spectrum method for
numerical simulation of free-space propagation in far and near fields,” Optics express,
vol. 17, no. 22, pp. 19662-19673, 2009.
[88] F. Shen, and A. Wang, “Fast-Fourier-transform based numerical integration method for
the Rayleigh-Sommerfeld diffraction formula,” Applied optics, vol. 45, no. 6, pp. 1102-
1110, 2006.
[89] W. L. Wong, X. Su, X. Li, C. M. G. Cheung, R. Klein, C.-Y. Cheng, and T. Y. Wong,
“Global prevalence of age-related macular degeneration and disease burden projection for
2020 and 2040: a systematic review and meta-analysis,” The Lancet Global Health, vol.
2, no. 2, pp. e106-e116, 2014.
[90] S. Ferrari, E. Di Iorio, V. Barbaro, D. Ponzin, F. S Sorrentino, and F. Parmeggiani,
“Retinitis pigmentosa: genes and disease mechanisms,” Current genomics, vol. 12, no. 4,
pp. 238-249, 2011.
[91] X. Qian, G. Lu, B. B. Thomas, R. Li, X. Chen, K. K. Shung, M. Humayun, and Q. Zhou,
“Noninvasive Ultrasound Retinal Stimulation for Vision Restoration at High
Spatiotemporal Resolution,” BME Frontiers, vol. 2022, 2022.
[92] H. Kang, X. Qian, R. Chen, R. Wodnicki, Y. Sun, R. Li, Y. Li, K. K. Shung, Z. Chen,
and Q. Zhou, “2-D ultrasonic array-based optical coherence elastography,” IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 68, no. 4, pp.
1096-1104, 2020.
[93] R. Wodnicki, H. Kang, R. Zhang, R. Chen, N. Cabrera-Munoz, J. Adhikari, C. Neher, C.-
T. Chiu, Y. Liu, and D. N. Stephens, "PIN-PMN-PT single crystal composite and 3D
printed interposer backing for ASIC integration of large aperture 2D array." pp. 1-4.
[94] B. B. Thomas, R. B. Aramant, S. R. Sadda, and M. J. Seiler, “Light response differences
in the superior colliculus of albino and pigmented rats,” Neuroscience letters, vol. 385,
no. 2, pp. 143-147, 2005.
[95] M. Tanter, and M. Fink, “Ultrafast imaging in biomedical ultrasound,” IEEE transactions
on ultrasonics, ferroelectrics, and frequency control, vol. 61, no. 1, pp. 102-119, 2014.
[96] E. Macé, G. Montaldo, I. Cohen, M. Baulac, M. Fink, and M. Tanter, “Functional
ultrasound imaging of the brain,” Nature methods, vol. 8, no. 8, pp. 662-664, 2011.
[97] H. Leinonen, and H. Tanila, “Vision in laboratory rodents—Tools to measure it and
implications for behavioral research,” Behavioural brain research, vol. 352, pp. 172-182,
2018.

78

[98] G. Li, W. Qiu, Z. Zhang, Q. Jiang, M. Su, R. Cai, Y. Li, F. Cai, Z. Deng, and D. Xu,
“Noninvasive ultrasonic neuromodulation in freely moving mice,” IEEE Transactions on
Biomedical Engineering, vol. 66, no. 1, pp. 217-224, 2018.
[99] Z. Zhang, W. Qiu, H. Gong, G. Li, Q. Jiang, P. Liang, H. Zheng, and P. Zhang, “Low-
intensity ultrasound suppresses low-Mg2+-induced epileptiform discharges in juvenile
mouse hippocampal slices,” Journal of neural engineering, vol. 16, no. 3, pp. 036006,
2019.
[100] D. Li, Z. Cui, S. Xu, T. Xu, S. Wu, A. Bouakaz, M. Wan, and S. Zhang, “Low-intensity
focused ultrasound stimulation treatment decreases blood pressure in spontaneously
hypertensive rats,” IEEE Transactions on Biomedical Engineering, 2020.
[101] M. Legrand, L. Galineau, A. Novell, B. Planchez, B. Brizard, S. Leman, C. Tauber, J.-M.
Escoffre, A. Lefèvre, and P. Gosset, “Efficacy of chronic ultrasound neurostimulation on
behaviors and distributed brain metabolism in depressive-like mice,” BioRxiv, pp.
813006, 2019.
[102] E. Bendau, C. Aurup, H. Kamimura, and E. Konofagou, “Low intensity, continuous wave
focused ultrasound reversibly depresses heart rate in anesthetized mice,” The Journal of
the Acoustical Society of America, vol. 146, no. 4, pp. 2942-2943, 2019.
[103] W. Lee, S. D. Lee, M. Y. Park, L. Foley, E. Purcell-Estabrook, H. Kim, K. Fischer, L.-S.
Maeng, and S.-S. Yoo, “Image-guided focused ultrasound-mediated regional brain
stimulation in sheep,” Ultrasound in medicine & biology, vol. 42, no. 2, pp. 459-470,
2016.
[104] L. Verhagen, C. Gallea, D. Folloni, C. Constans, D. E. Jensen, H. Ahnine, L.
Roumazeilles, M. Santin, B. Ahmed, and S. Lehéricy, “Offline impact of transcranial
focused ultrasound on cortical activation in primates,” Elife, vol. 8, pp. e40541, 2019.
[105] H. A. Kamimura, S. A. Lee, Y. Niimi, C. Aurup, M. G. Kim, and E. E. Konofagou,
"Focused ultrasound stimulation of median nerve modulates somatosensory evoked
responses." pp. 1085-1087.
[106] J. Kubanek, “Neuromodulation with transcranial focused ultrasound,” Neurosurgical
focus, vol. 44, no. 2, pp. E14, 2018.
[107] J. Mueller, W. Legon, A. Opitz, T. F. Sato, and W. J. Tyler, “Transcranial focused
ultrasound modulates intrinsic and evoked EEG dynamics,” Brain stimulation, vol. 7, no.
6, pp. 900-908, 2014.
[108] W. M. Grill, “Safety considerations for deep brain stimulation: review and analysis,”
Expert review of medical devices, vol. 2, no. 4, pp. 409-420, 2005.
[109] A. Amon, and F. Alesch, “Systems for deep brain stimulation: review of technical
features,” Journal of Neural Transmission, vol. 124, no. 9, pp. 1083-1091, 2017.

79

[110] H. Imai, S. Honda, Y. Nakanishi, H. Yamamoto, Y. Tsukahara, and A. Negi, “Different
transitions of multifocal electroretinogram recordings between patients with age-related
macular degeneration and polypoidal choroidal vasculopathy after photodynamic
therapy,” British journal of ophthalmology, vol. 90, no. 12, pp. 1524-1530, 2006.
[111] C. A. Curcio, N. E. Medeiros, and C. L. Millican, “Photoreceptor loss in age-related
macular degeneration,” Investigative ophthalmology & visual science, vol. 37, no. 7, pp.
1236-1249, 1996.
[112] S. B. Ryu, P. Werginz, and S. I. Fried, “Response of Mouse Visual Cortical Neurons to
Electric Stimulation of the Retina,” Frontiers in neuroscience, vol. 13, 2019.
[113] H. Kim, M. Y. Park, S. D. Lee, W. Lee, A. Chiu, and S.-S. Yoo, “Suppression of EEG
visual-evoked potentials in rats via neuromodulatory focused ultrasound,” Neuroreport,
vol. 26, no. 4, pp. 211, 2015.
[114] G. Menikou, and C. Damianou, “Acoustic and thermal characterization of agar based
phantoms used for evaluating focused ultrasound exposures,” Journal of therapeutic
ultrasound, vol. 5, no. 1, pp. 14, 2017.
[115] P. Gabbott, and M. Stewart, “Distribution of neurons and glia in the visual cortex (area
17) of the adult albino rat: a quantitative description,” Neuroscience, vol. 21, no. 3, pp.
833-845, 1987.
[116] M. Mohammadjavadi, P. P. Ye, A. Xia, J. Brown, G. Popelka, and K. B. Pauly,
“Elimination of peripheral auditory pathway activation does not affect motor responses
from ultrasound neuromodulation,” Brain stimulation, vol. 12, no. 4, pp. 901-910, 2019.
[117] M. A. O’Reilly, A. Muller, and K. Hynynen, “Ultrasound insertion loss of rat parietal
bone appears to be proportional to animal mass at submegahertz frequencies,”
Ultrasound in medicine & biology, vol. 37, no. 11, pp. 1930-1937, 2011.
[118] M. Gerstenmayer, B. Fellah, R. Magnin, E. Selingue, and B. Larrat, “Acoustic
transmission factor through the rat skull as a function of body mass, frequency and
position,” Ultrasound in medicine & biology, vol. 44, no. 11, pp. 2336-2344, 2018.
[119] K. Hynynen, N. I. Vykhodtseva, A. H. Chung, V. Sorrentino, V. Colucci, and F. A.
Jolesz, “Thermal effects of focused ultrasound on the brain: determination with MR
imaging,” Radiology, vol. 204, no. 1, pp. 247-253, 1997.
[120] O. Rudenko, A. Sarvazyan, and S. Y. Emelianov, “Acoustic radiation force and
streaming induced by focused nonlinear ultrasound in a dissipative medium,” The
Journal of the Acoustical Society of America, vol. 99, no. 5, pp. 2791-2798, 1996.
[121] W. J. Tyler, S. W. Lani, and G. M. Hwang, “Ultrasonic modulation of neural circuit
activity,” Current opinion in neurobiology, vol. 50, pp. 222-231, 2018.

80

[122] M. Plaksin, S. Shoham, and E. Kimmel, “Intramembrane cavitation as a predictive bio-
piezoelectric mechanism for ultrasonic brain stimulation,” Physical review X, vol. 4, no.
1, pp. 011004, 2014.
[123] B. Krasovitski, V. Frenkel, S. Shoham, and E. Kimmel, “Intramembrane cavitation as a
unifying mechanism for ultrasound-induced bioeffects,” Proceedings of the National
Academy of Sciences, vol. 108, no. 8, pp. 3258-3263, 2011.

Abstract (if available)

Abstract Neuromodulation is a technology that directly stimulates neurons and modulates neural activities through the targeted delivery of a stimulus to specific neurological sites in the body. Since the first clinical success of the use of electrical stimulation for pain relief, neuromodulation has been widely employed in prostheses or neuropathic disease treatment.
Electrical and chemical stimulations are the most conventional techniques for neuromodulation. They are widely used in clinics and research, but they also have limitations, such as invasiveness and resolution. In recent decades, different types of new neuromodulation technologies have been investigated, including transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), optogenetics, sonogenetics, and ultrasound (US) stimulation. Of these emerging methods, US stimulation has unique advantages, including noninvasiveness, deep penetration, high resolution, and cost-effectiveness.
The work presented in this dissertation investigates the feasibility of US stimulation for visual prostheses. It was demonstrated that focused US stimulation could evoke neural activities in the retina and visual cortex. Experiments were conducted in vivo on normal-sighted Long Evans rats and retinal degenerative Royal College of Surgeons rats. Parameters of US such as amplitude and duration were optimized for stimulation by quantifying the evoked neural responses. The spatiotemporal resolution of US stimulation was investigated by mapping the neuron response with a multi-electrode array (MEA) and changing US waveforms. A 2D array-based US stimulation with arbitrary patterns was also presented. Water-licking tests were conducted to demonstrate that animals’ behavioral responses to US retinal stimulation are similar to those of light stimulation. In addition, the safety and physical mechanisms of US stimulations were investigated by measuring the amplitude of US pressure and the US-induced heating effect. Eye imaging and histology were also carried out to ensure there was no morphological damage to the stimulated area. The results indicated the capability of the US stimulation to evoke neuronal activity. US-based visual prosthesis could be a promising technology for blindness treatment. Furthermore, the favorable prospect of the ultrasonic retinal prosthesis in the next step is predicted. Our future study will also examine the biological mechanism of US stimulation via in vitro and single-cell stimulation.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Therapeutic electrical stimulation strategies for neuroregeneration and neuroprotection of retinal neurons

PDF

Manipulation of RGCs response using different stimulation strategies for retinal prosthesis

PDF

Stimulation strategies to improve efficiency and temporal resolution of epiretinal prostheses

PDF

Next generation neural interfaces for neural modulation and prosthesis

PDF

Design and application of a C-shaped miniaturized coil for transcranial magnetic stimulation in rodents

PDF

Towards a high resolution retinal implant

PDF

PET study of retinal prosthesis functionality

PDF

Improving stimulation strategies for epiretinal prostheses

PDF

Prosthetic vision in blind human patients: Predicting the percepts of epiretinal stimulation

PDF

Effect of continuous electrical stimulation on retinal structure and function

PDF

High frequency ultrasound elastography and its biomedical applications

PDF

Synaptic second wind: phosphene fading of the Argus II and glutamate secretion in the inner retina

PDF

Characterization of visual cortex function in late-blind individuals with retinitis pigmentosa and Argus II patients

PDF

Cortical and subcortical responses to electrical stimulation of rat retina

PDF

High resolution elastography in ophthalmology

PDF

On the electrophysiology of multielectrode recordings of the basal ganglia and thalamus to improve DBS therapy for children with secondary dystonia

PDF

Configurable imaging platform for super-harmonic contrast-enhanced ultrasound imaging

PDF

Simulating electrical stimulation and recording in a multi-scale model of the hippocampus

PDF

2D ultrasonic transducer array’s design and fabrication with 3D printed interposer and applications

PDF

Circuits for Biomedical Telemetry and Stimulation Applications

Asset Metadata

Creator Lu, Gengxi "Alex" (author)

Core Title Ultrasound neuromodulation and its applications for noninvasive vision restoration

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Degree Conferral Date 2022-12

Publication Date 05/14/2023

Defense Date 10/20/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag blindness,medical imaging,noninvasive neuromodulation,OAI-PMH Harvest,retina prosthesis,ultrasound stimulation

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Zhou, Qifa (committee chair), Humayun, Mark (committee member), Khoo, Michael (committee member)

Creator Email gengxilu@usc.edu,loadinglgx@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112417838

Unique identifier UC112417838

Identifier etd-LuGengxiAl-11313.pdf (filename)

Legacy Identifier etd-LuGengxiAl-11313

Document Type Dissertation

Format theses (aat)

Rights Lu, Gengxi "Alex"

Internet Media Type application/pdf

Type texts

Source 20221115-usctheses-batch-991 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu