University of Southern California Dissertations and Theses

High frequency ultrasonic imaging for the development of breast biopsy needle with miniature ultrasonic transducer array

(USC Thesis Other)

High frequency ultrasonic imaging for the development of breast biopsy needle with miniature ultrasonic transducer array

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content HIGHFREQUENCYULTRASONICIMAGINGFORTHEDEVELOPMENTOFBREAST BIOPSYNEEDLEWITHMINIATUREULTRASONICTRANSDUCERARRAY by TakahiroAkiyama AThesisPresentedtothe FACULTYOFTHEUSCGRADUATESCHOOL UNIVERSITYOFSOUTHERNCALIFORNIA InPartialFulfillmentsofthe RequirementsfortheDegree MASTEROFSCIENCE (BIOMEDICALENGINEERING) May2016 Copyright2016 TakahiroAkiyama Acknowledgement IamexceedinglygratefulforthesupportoftheNIHResourceCenterforMedicalUltrasonicTransducer TechnologyattheUniversityofSouthernCaliforniawhichhasprovidedmewiththegreatopportunityto studyandperformresearchonultrasoundandhasalsoallowedmetocollaboratewithwonderfulfaculty members and colleagues. I am also sincerely thankful to Canon Inc. for financial support, including tuitionandlivingexpenses,whichhasmadepossibletheachievementsthathavebeenrealizedhere. Firstly, I would like to profoundly thank Dr. K. Kirk Shung for introducing me to the ultrasound world as well as for his guidance and patience during my graduate studies. Dr. Shung accepted a much more complicated situation with me than with other students. Moreover, Dr. Shung supported me to buildupanewtopicareaandintroducedmetoamedicaldoctorforcollaboration. Furthermore, Dr, Shung’s mentorship is superb, and provided sophisticated knowledge and experi- ence,whichfulfilledmystudylifehereandassistedmeintheattainmentofmystudy. I would like to acknowledge Dr. Sue E. Martin in the Keck School of Medicine at the University of SouthernCaliforniaforgivingmepreciouscommentsaboutbreastpathology. I further wish to thank Dr. Jesse T. Yen and Dr. Mike S. W. Chen for being a part of my thesis committee. Despite their busy schedules as professors, they supported my achievement of the thesis. I wouldliketoshowmyappreciationforDr. QifaZhou,whofrequentlyadvisedmeaboutmycareerand studyinthelaboratory,whichwashelpfulformyprojectandmyfuturecareer. i I am thankful to Ms. Mischalgrace Diasanta for her sincerity during the duration of my study pro- gram. When I faced some difficulties about the curriculum of biomedical engineering, she helped me everytime. Herparamountskillsasanadvisorledmysuccessinthisprogram. IwouldalsoliketoexpressmythankstoDr. ThomasCummins,Dr. BongJinKang,Dr. ChangYang Lee, Mr. Nester E. Cabrera-Munoz, Mr. Chi Tat Chiu, Mr. Robert Wodnicki, Mr. Payam Eliahoo, Mr. ZeyuChen,Dr. ChunlongFeiandothercolleaguesinUTRCfortheirkindness,supportandfriendship. Dr. Kyle D. Squires who is a professor at Arizona State University, Tempe, AZ, Dr. Koichi Hishida whoisaprofessoratKeioUniversity,Tokyo,Japan,andDr. YutakaWatanabewhowasaseniorgeneral researcher of Canon Inc., Tokyo, Japan, recommended me to enter the USC Master’s degree program andtostudybiomedicalengineering. Iwouldliketothankthemfortheirassistanceinmystudyandlife. Dr. Toshiaki Ikoma, who is a former vise president and a former CTO of Canon Inc., suggested the enlargementoftheglobalstudyprograminmycompany,andsupportedmyparticipationinthisprogram. I would like to sincerely thank him because I couldn’t have had any precious experiences here without hisenthusiasmfortheimprovementofresearchanddevelopmentformycompany. IamgratefultoMr. HideshiKawasakiandMr. KazutoshiTorashimainCanonInc.,fortheirsupport oftheprogram. Ifrequentlyconsultedwiththemaboutthecontents,planandmylifeheresothatIcould accomplishmygoalofthisprogramwithoutanygreatconcerns. Iamalsothankfultomycolleaguesin mycompany. Lastbutnotleast,Iwouldliketodeeplythankmyamazingwife,YokoAkiyama,andmyfamilyfor theirbenevolentassistanceinmystudyandlife. TakahiroAkiyama ii Contents 1 Introduction 1 1.1 MedicalUltrasound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 BreastCancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.3 ObjectiveoftheResearch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2 BreastBiopsyTissue 28 2.1 BreastBiopsyTissue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.1 Breastcellsandtissues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.2 Methodofpathology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.1.3 HistologicalImageofBreastTissue . . . . . . . . . . . . . . . . . . . . . . . . 35 3 UltrasonicTransducerandImaging 37 3.1 UltrasonicTransducer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 3.1.1 SingleElementUltrasonicTransducer . . . . . . . . . . . . . . . . . . . . . . . 39 3.1.2 LinearArrayUltrasonicTransducer . . . . . . . . . . . . . . . . . . . . . . . . 42 3.2 UltrasonicImaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.2.1 B-mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.2.2 SyntheticApertureImaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.2.3 BackscatteringAnalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.2.4 StatisticalAnalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4 QuantitativeBackscatteringAnalysisforCharacterizationofBreastBiopsyTissue 68 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 4.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 4.2.1 TissueSpecimenAcquisitionandProcessing . . . . . . . . . . . . . . . . . . . 70 4.2.2 TissueSpecimenImagingExperimentalSetup . . . . . . . . . . . . . . . . . . 72 4.2.3 QuantitativeBackscatteringAnalysis . . . . . . . . . . . . . . . . . . . . . . . 77 4.2.4 QuantitativeandStatisticalAnalysis . . . . . . . . . . . . . . . . . . . . . . . . 79 4.3 ResultandDiscussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 iii 4.3.1 2DMapsofBackscatteringCoefficient . . . . . . . . . . . . . . . . . . . . . . 82 4.3.2 StatisticalAnalysisResults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 4.3.3 AdditionalDiscussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 5 BuildingTestingSystemforHighFrequencyUltrasonicTransducerArray 103 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 5.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 5.2.1 SyntheticApertureImagingSystem . . . . . . . . . . . . . . . . . . . . . . . . 105 5.2.2 SoftwareforSyntheticApertureImaging . . . . . . . . . . . . . . . . . . . . . 105 5.2.3 MiniatureUltrasonicTransducerArray . . . . . . . . . . . . . . . . . . . . . . 110 5.3 ResultandDiscussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 6 Conclusion 120 6.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 6.2 FutureWork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Reference 123 iv ListofTables 1.1 Anexampleoftransducerguide,courtesyofGEhealthcare . . . . . . . . . . . . . . . 7 1.2 ComparisonofImagingModalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.3 Stagesofbreastcancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.4 Breast cancer statistics in the U.S. and Japan, the numbers are population of new cases anddeathinayearandthenumbersinsideofparenthesesaretheratesper100,000ofeach. 18 4.1 Thecharacteristicsofthesingleelementultrasonictransducerusedtoacquireultrasound echofromthespecimensofbreastbiopsy. . . . . . . . . . . . . . . . . . . . . . . . . . 75 4.2 Thebasicinformationofthespecimens. . . . . . . . . . . . . . . . . . . . . . . . . . . 82 4.3 Meanandstandarddeviationofbackscatteringparametersincase1. . . . . . . . . . . . 83 4.4 Meanandstandarddeviationofbackscatteringparametersincase2. . . . . . . . . . . . 85 4.5 Meanandstandarddeviationofbackscatteringparametersincase3. . . . . . . . . . . . 87 4.6 Meanandstandarddeviationofbackscatteringparametersincase4. . . . . . . . . . . . 89 4.7 Mean and standard deviation of backscattering parameters in case 1, which is compen- satedbyusingtheattenuationcoefficientextractedfromtheradiofrequencyechosignal. 98 4.8 Meanandstandarddeviationofbackscatteringparametersofadipose,whichiscompen- satedbyusingtheattenuationcoefficient. . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.9 Mean and standard deviation of backscattering parameters of adenocarcinoma, which is compensatedbyusingtheattenuationcoefficient. . . . . . . . . . . . . . . . . . . . . . 100 5.1 ScantypesinSyntheticApertureImaging. . . . . . . . . . . . . . . . . . . . . . . . . . 109 5.2 Acousticstackproperties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 5.3 Themeasurementspecificationoftheminiaturetransducer. . . . . . . . . . . . . . . . . 112 v ListofFigures 1.1 Ultrasoundprobes. CourtesyofGEHealthcare. . . . . . . . . . . . . . . . . . . . . . . 6 1.2 UltrasonicTransducer,(a)singleelementtransducer,courtesyofOlympusand(b)linear array,courtesyofIntelligenceofmedicaltechnologies. . . . . . . . . . . . . . . . . . . 8 1.3 Field of view of 2-D images corresponding to (a) linear array, (b) convex array, and (c) phasedarray. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.4 Imaging platform developed in Ultrasonic Transducer Resource Center at University of SouthernCalifornia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.5 Imagingplatformforresearch,courtesyofVerasonics. . . . . . . . . . . . . . . . . . . 11 1.6 Intravascularultrasound,courtesyofVolcano. . . . . . . . . . . . . . . . . . . . . . . . 12 1.7 Breastanatomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.8 Lactiferousduct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.9 Ductal carcinoma in Situ, showing how ductal carcinoma in situ develops from non- invasivetoinvasivecarcinoma[20]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 1.10 Invasiveductalcarcinoma. A:Ducts,B:Lobules,C:Dilatedsectionofducttoholdmilk, D:Nipple,E:Fat,F:PectoralismajormuscleG:Chestwall/ribcage. Inenlargement,A; Normal duct cells, B: Ductal cancer cells breaking through the basement membrane, C: Basementmembrane[20]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 1.11 Positionofultrasoundprobeinbreastultrasound[19]. . . . . . . . . . . . . . . . . . . . 21 1.12 Normalbreastultrasoundandgraphic[19]. . . . . . . . . . . . . . . . . . . . . . . . . 22 1.13 Mammography,courtesyofUniversityofConnecticutHealthCenter.. . . . . . . . . . . 23 1.14 Standardimagesobtainedforscreeningmammography[19]. . . . . . . . . . . . . . . . 23 1.15 Fineneedleaspirationbiopsyofthebreast[19]. . . . . . . . . . . . . . . . . . . . . . . 24 1.16 Vacuum-assistedbiopsyofthebreast[19]. . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.17 Prefirelocationandpostfirelocationofultrasound-guidedneedlebiopsy[18]. . . . . . . 25 1.18 A trial piece of a breast biopsy needle with a miniature high frequency ultrasonic trans- ducerarray[15].. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 1.19 The proposed concept of ultrasound-guided breast biopsy combining low and high fre- quencyultrasound. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 vi 2.1 Microcalcifications. Artificialdisruptionoftissue[30]. . . . . . . . . . . . . . . . . . . 31 2.2 Mammogramofmicrocalcifications. [18]. . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.3 Adenocarcinoma,10timesmagnification. [31] . . . . . . . . . . . . . . . . . . . . . . 32 2.4 Core needle breast biopsy [18]. (a) Automated spring-loaded core needle tip in the un- fired (top) and fired (bottom) position . (b) Size of a 16-gauge samples versus 14-gauge samplesobtainedwithanautomatedspring-loadedcoredevice. . . . . . . . . . . . . . 33 2.5 Sectionsbeingcutfromaparaffinblockusingarotarymicrotome. CourtesyofLeica. . . 35 2.6 Pathologymicroscopeslides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.7 Normalbreast,glandsandstroma. CourtesyofDr. ShashidharinJamesCookUniversity, Australia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 2.8 Fibroadenomathatisbenign. Intracanaliculargrowth. [30]. . . . . . . . . . . . . . . . . 36 2.9 Ductalcarcinomainsitu(DCIS)ofcribriformpattern. Multiplespaceswithintheprolif- erationofmonotonouscellsareroundedanddistributedinanorganizedfashion[29]. . . 36 2.10 Invasiveductalcarcinoma,welldifferentiated(gradeI)[29]. . . . . . . . . . . . . . . . 36 3.1 Piezoelectric transducer element geometry for (a) a single element and for (b) a linear array. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.2 Aphotographofsingleelementultrasonictransducer. . . . . . . . . . . . . . . . . . . . 41 3.3 Schematicsoflens-focusedandpress-focusedtransducers. . . . . . . . . . . . . . . . . 41 3.4 UBMforopthalmology. CourtesyofQuantelMedical,France. . . . . . . . . . . . . . . 42 3.5 UBMfordermatology. CourtesyofDermascan,Denmark. . . . . . . . . . . . . . . . . 42 3.6 Constructionofaone-dimensionalarraywithanelevationplanelens. [34] . . . . . . . . 43 3.7 Ultrasound array technology progression with dice-and-fill, laser micromachining and DRIEmicromachiningfabricationmethods. [36] . . . . . . . . . . . . . . . . . . . . . 44 3.8 Structureof30MHzcompositearray[35]. . . . . . . . . . . . . . . . . . . . . . . . . . 45 3.9 Photographof30MHzlineararrayassembly[13]. . . . . . . . . . . . . . . . . . . . . . 45 3.10 Singleelementmechanicalsectorscannerandlineararrayscanner. [1] . . . . . . . . . . 46 3.11 BlockdiagramofasingleelementB-modescanner. [1] . . . . . . . . . . . . . . . . . . 52 3.12 SchematicofB-modescanningbyusingasingleelementultrasonictransducer,theimage isoriginalfromaclassofBME535orBME536atUniversityofSouthernCalifornia. . . 53 3.13 Theconceptofreceivefocusbylineararray . . . . . . . . . . . . . . . . . . . . . . . . 54 3.14 Block diagram of an ultrasonic imaging system architecture that utilizes a digital beam former. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.15 Highresolvedimagecanbereconstructedbysummationofeachlowresolvedimagethat isgeneratedbyoneelementtransmission. [38]. . . . . . . . . . . . . . . . . . . . . . . 56 3.16 Geometric relation between the transmit and receive element combination and the focal point. [38] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.17 Spectral analysis procedure of backscattering coefficient. (a) Section of B-mode image, (b)logofbackscatteringcoefficientmagnitudevsfrequency. [52] . . . . . . . . . . . . 61 3.18 Flowchartofstatisticalanalysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 vii 4.1 Outline of clinical study to prove the efficacy of high frequency ultrasound imaging in identifyingcancerinbreastbiopsyspecimens. . . . . . . . . . . . . . . . . . . . . . . . 71 4.2 The illustration of experimental apparatus for the acquisition of ultrasound echo from breastbiopsytissuespecimensusingsingleelementhighfrequencyultrasonictransducer. 72 4.3 Biopsyspecimensweresecuredinanagargelblocktomaintaintheirorientationduring ultrasoundandhistologicalimaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 4.4 ThephotooftheexperimentalarrangementofUBMonthemeasurementofpulseecho. . 74 4.5 The data of the pulse echo of the single element ultrasonic transducer used for the backscattering analysis: the left shows time response of the pulse echo and the right showsfrequencyspectrumindBbasedonthepeakvalue. . . . . . . . . . . . . . . . . . 75 4.6 The comparison of B-mode images: the top shows low frequency components are cut off and the bottom is without application of the filter. The bottom one shows relatively apparentspecklesindeeperareathatindicatesfattytissue,onthecontrary,itshowsmuch opaqueinthetopone. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 4.7 Thetendencyoftheresultofbackscatteringcoefficientinwindowlength. . . . . . . . . 77 4.8 Theflowchartofthebackscatteringanalysis. . . . . . . . . . . . . . . . . . . . . . . . 78 4.9 The example of setting the region of interest for statistical analysis of backscattering coefficients. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 4.10 Case 1 image set includes high frequency ultrasound. The red dashed circles indicate adiposetissueareasandgreencirclesindicateadenocarcinoma. . . . . . . . . . . . . . . 84 4.11 Case2imagesets: leftrowispart1andrightrowispart2, includehighfrequencyultra- sound. Theredarrowsindicatemicrocalcificationsandtheredandgreencirclesindicate fibroustissueandadiposetissuerespectively. . . . . . . . . . . . . . . . . . . . . . . . 86 4.12 Case 3 image set includes high frequency ultrasound. The histological image shows the biopsyspecimencomposesofadenocarcinoma. . . . . . . . . . . . . . . . . . . . . . . 88 4.13 Case 4 image set includes high frequency ultrasound. The red arrows indicate the posi- tionsofmicrocalcifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 4.14 Box plots for integrated backscattering coefficient (top), slope (middle), and y-intercept (bottom)ofadiposeandadenocarcinomaincase1. . . . . . . . . . . . . . . . . . . . . 92 4.15 Box plots for integrated backscattering coefficient (top), slope (middle), and y-intercept (bottom)ofadipose,fibroustissueandmicrocalcificationsincase2. . . . . . . . . . . . 94 4.16 Box plots for integrated backscattering coefficient (left), slope (center), and y-intercept (right)ofadenocarcinomaincase3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.17 Box plots for integrated backscattering coefficient (top), slope (middle), and y-intercept (bottom)offibroustissueandadenocarcinomaincase4. . . . . . . . . . . . . . . . . . 96 4.18 Attenuationcoefficientincase1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 4.19 Attenuationcoefficientsofseveralkindsoftissues . . . . . . . . . . . . . . . . . . . . . 99 4.20 Beamprofileofthesingleelementtransducerusedinthisstudy. Leftimageis2Dprofile ofthepressuregeneratedbythetransducer,rightimageshowstheaxialprofile. . . . . . 101 viii 5.1 Imagingsystemblockdiagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 5.2 Themultiplexerprovidesindependentswitchingbetweentransmitandreceivechannels. 107 5.3 GraphicuserinterfaceprogrammedinLabVIEWforsyntheticapertureimaging. . . . . 108 5.4 Transducerontheglasswithtransmissionlines. . . . . . . . . . . . . . . . . . . . . . . 110 5.5 Zoomedviewofthetransducer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 5.6 Anexplodedarrayviewoftheminiatureultrasonictransducer. . . . . . . . . . . . . . . 113 5.7 ScanningelectronmicroscopeimageofPMN-PTmaterialdirectlyafterDRIEprocessing.113 5.8 Pulse-echo measured results for a typical array element: (left) echo signal in time and (right)thatinfrequencyindB. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 5.9 Centerfrequencyandbandwidthvaluesforeachelementinarray. . . . . . . . . . . . . 114 5.10 Experimentalconfigurationoftheneedlearrayforimagingawirephantom. . . . . . . . 114 5.11 Reconstructed synthetic aperture image of a single 20 mm wire target. The B-mode imagewasdisplayedonalineargrayscaleindBrelativetomaximumintensity. . . . . . 116 5.12 Axialandlaterallineplotsforthecenterofthewirephantom. . . . . . . . . . . . . . . 117 ix Abstract High frequency ultrasonic transducers integrated in a breast biopsy needle have been researched at the University of Southern California. It can obtain information in the vicinity of the needle and resolve microcalcificationsthatareindicativeofprecancerouschangesorearlybreastcarcinomassuchasductal carcinoma in situ. X-ray mammography is the best diagnostic modality to detect microcalcifications currently, but patients are exposed to ionizing radiation. More recently, a new ultrasound-guided breast biopsy technique was proposed. This technique utilizes conventional ultrasound guidance coupled with high frequency ultrasound from the biopsy needle. Current ultrasound-guided breast biopsy can often result in false negative diagnosis that causes serious consequences because of the inaccuracy of the position of the biopsy needle. This proposed technology can improve localization of lesions and the accuracyofthediagnosisofbreastpathology. Inthisstudy,forthepurposeofdemonstratingtheproofoftheconceptwithrespecttotheminiature high frequency ultrasonic transducer embedded within a biopsy needle, backscattering analysis for the characterization of breast biopsy tissues was conducted by using a single element high frequency ul- trasonic transducer with a center frequency of 74 MHz. The effective bandwidth was from 41 MHz to 88 MHz and was defined on - 8 dB relative to the maximum value of the pulse echo spectrum on the transducer. The slope and the 0 Hz y-intercept of the spectral backscattering features were calculated with linear approximation as well as the integrated backscattering coefficient. Four breast x biopsy tissue specimens in each agar block used for embedding were measured utilizing a standard ultrasonic bio-microscopy (UBM) scanning methodology. All of the backscattering coefficients were two-dimensionallymappedandco-registeredwitheachhistopathologicalimagefortheidentificationof tissuetypes. The results of the analysis indicate that the integrated backscattering coefficient successfully differ- entiated microcalcifications from other tissues around them. And adenocarcinoma is also differentiated fromadiposetissuesbyobservingtheintegratedbackscatteringandtheslopeofthespectralbackscatter- ingfeature. Theseresultsindicatethatbackscatteringanalysisisabletoquantitativelydistinguishtissues into normal and abnormal, which should help radiologists locate abnormal areas during the proposed ultrasound-guidedbreastbiopsywithhighfrequencyultrasound. Currentimagingsystemsforconventionalultrasoundareavailableforlessthan20MHz. Toevaluate the miniature high frequency ultrasonic transducer, a test system with synthetic aperture technique was built , and used to process the data to produce ultrasound images without any large scale electronics. TheGUIcontrollingthedataacquisitionandtheimageprocessingprogramarebeneficialforthosewho research high frequency ultrasonic transducer arrays. It takes a few minutes to reconstruct a synthetic B-modeimageaftertheacquisitionoftheRFechosignals. Inconclusion,thebackscatteringanalysisshowsthathighfrequencyultrasoundissignificantlyuseful for the proposed ultrasound-guided breast biopsy, and the testing system based on a GUI should be flexiblyavailablefortheevaluationofeveryhighfrequencytransducerarray. Therefore,thedevelopment of the miniature high frequency ultrasonic transducer integrated in a breast biopsy needle can be easily advancedtothenextstage. xi Chapter1 Introduction 1 1.1 MedicalUltrasound 1.1 MedicalUltrasound During World War II, the technologies which would be fundamental of ultrasound were developed in a number of countries including Japan and the United States. After World War II, ultrasound for medical diagnosis started to evolve in these countries. In 1950’s, a prototype B-mode ultrasonic imaging was developed in the United States, demonstrating the capability of ultrasound to characterize cancerous tissue. After that, on the opposite side of the pacific ocean, Doppler technology which had the potential to observe the movement of tissues and blood flow was invented in Japan. Ultrasound was not widely usedasastandardimagingmodalitylikeX-rayradiographyuntilearly1970’swhenthegray-scalelevel mapping that everybody has seen today was introduced. Doppler flow meters to detect blood flow also became available during that time, which made ultrasound practical for the imaging of anatomy and the measurement of blood flow. Thus, ultrasound is one of the most conventional and standardized diagnostictoolsinmedicineandutilizedasasignificantimagingmodalityatanymedicalfacilitiesinthe worldtoday[1],[2]. The ultrasound technology is advancing at a rapid pace even today. Undoubtedly, the development ofelectronics,informationtechnology,manufacturing,materialscience,collaborationbetweenmedicine and engineering also played important roles. The real-time system that exhibited good image quality of ultrasound did not appear until 1981 [3]. Programable integrated circuit technology including micro- processoranddigitalmemoryenabledcommercialreal-timephased-arraysystems. Application-specific integratedcircuit(ASICs),digitalsignalprocessingchips(DSPs),andthecomputer-aideddesign(CAD) of very large scale integration (VLSI) circuits came along as well during 1980’s, which were benefi- cial for the array system to be the dominant imaging modality. The development of electronics also contributed to realizing the system viewing blood flow in real time by using color Doppler imaging technology. In parallel with the innovation of the systems, ultrasonic transducers were also rapidly de- veloped. TheimprovementofequivalentcircuitmodelbasedonMasonmodelledtospecializedphased 2 1.1 MedicalUltrasound and linear arrays for specific clinical application such as cardiology, radiology, endoscopic, transvagi- nal, and so forth. Arrays with several hundred elements and higher frequencies were realized by the discovery and the improvement of materials and piezoelectric composites. Thanks to the progress of piezoelectric material, wider bandwidth transducer became available and able to transmit ultrasound at severalfrequenciesselectablebytheoperatorwithouthandlingmanyprobes[2]. Contrast agents assisted the growth of ultrasound imaging extensively. They were designed to en- hance the visibility of flowing blood. Microbubbles are one kind of the contrast agents which respond totheultrasoundfromatransducerandemitotherfrequencythanfundamentalfrequencythatthetrans- ducer transmits. Manufacturers designed the contrast agents to respond wave at harmonic frequencies in order to visualize the image with high contrast by filtering fundamental echo. Interestingly, on the processofthedevelopmentofharmonicimagingbyusingcontrastagents,tissueharmonicimagingwas accidentally discovered. By the late of 1990’s, the tissue harmonic imaging without contrast agents becamerecognizedasindispensabletoolinultrasound. [2]. Further innovation of electronics occurred in 1990’s introduced more powerful microprocessors, high-density gate arrays and surface-mount technology and low-cost A/D convertors that enabled the ultrasound imaging systems to be smaller in volume and faster in processing with evolving digital ar- chitectures and beamformers. Transducers with much wider bandwidth, 1.5D: elevation focusing, and matrix arrays were developed because the innovative electronics had the capability of processing sig- nals from numerous elements of the transducer. Nowadays, 3D real-time imaging by using 2D array transducerhavebeenintroduced. Recently, other ultrasound application has been successively developed. Elastography which can detect elastic and viscoelastic feature of target tissues was commercialized as the tool for diagnosing lesions that have varied elasticity [4]. For example, tumors in liver may be slightly harder than healthy tissuesaroundthetumors. AlthoughB-modeiswidelyeffectiveforimagingorgans,itsometimescannot distinguish the tumors from healthy tissues. Elastography, on the contrary, can differentiate them by 3 1.1 MedicalUltrasound propagating ultrasound longitudinal and shear wave. It is eventually recognized as one of the valuable toolsforcharacterizingmanykindsoftumorswhichcanbeinvisibleonlywhenusingconventionalultra- sound. Anothertissuecharacterizationmethodisbackscatteringanalysisnotusingenvelopedsignalbut radio frequency echo signal [5]. Especially, integrated backscattering coefficients are already applied to intravascularultrasound(IVUS)todistinguishplaquepathologythatexistsonthewallofcoronaryartery [6] . Volcano (San Diego, CA) is a representative company that has succeeded producing commercial IVUSsystemswiththefunctionofthecharacterizationoftheplaque. Combining modalities can creates novel diagnostic methods. Image fusion is the combination of different image modalities, viewing images such as MRI, CT and ultrasound [7]. Compensating each other, it renders a lot of information which can be significant for diagnosis. Multi-wave interactive imagingisalsoanothercomplementaryways. Oneofthemiselastographyespeciallyacousticradiation force imaging (ARFI) that induces displacement by stronger pulses [8]. The displacement is dependent on the stiffness of the tissue so that an image of displacements can be formed. Another example is photoacoustic imaging. Photoacoustics is the phenomenon which creates acoustic displacement and waves caused by light exposing on the tissue. The acoustic waves are sensed by a conventional array and displayed on an imaging system. The amplitude and frequency are dependent on the characteristic ofthetissueonwhichthelightimpingessuchasthesize,heatcapacitance,compressibilityofthetissue; therebyphotoacousticspresentsadifferentimagefromaconventionalgray-scaleultrasoundimage[9]. Ultrasoundisutilizednotonlyforimagingbuttreatment. Highintensityfocusedultrasound(HIFU) isoneoftheapplicationsofitwhichisbasedonthehighlyfocusedultrasoundbeamtargetingaspecific part of the body such as tumors. These innovations of ultrasound application have further enhanced the capabilityofultrasoundinthemedicalfield. Conventionalultrasoundiswidelyusedformedicalimagingmodalitiesduetoitssafety,portability, costeffectiveandrealtimecapabilities,usingfrequencyrangesfrom2MHzto20MHzandspatialres- olutionisontheorderofafewmillimeters. Asvaluablediagnostictools,theyareprevailedandutilized 4 1.1 MedicalUltrasound inmanymedicaldisciplines. Ultrasonictransducerswhichtransmitandreceiveacousticwavesareindis- pensablefortheultrasonicimagingtechnology. Figure1.1showsprobesforthemedicalultrasound. Most of the transducers are made of piezoelectric materials which can convert the energy between electrical and mechanical including acoustic energy. Although electrostatic transducers have been researched and developed[10],[11],[12],thetransducerofpiezoelectricmaterialiscurrentlymainstream. Theprinciple ofultrasonicimagingistotransmitultrasoundwavefromthetransducersbyappliedelectricalpulsesto it and to receive the echo from tissues due to the mismatching acoustic impedance, converted to corre- sponding electrical signals, which can be amplified and processed to reconstruct images. The strength of the echo reflected and scattered at the points of the tissue of interest is dependent on the acoustically inhomogenous structure of the tissue. Therefore, the echo carries valuable information about the tissue and tell us about them. To acquire two dimensional image, the acoustic beam from the transducer is scannedazimuthally orlaterally over theregion of interest is necessary. The scanlines frequently called A-linesobtainedareusedtoconstructa2Dimage. Theechoamplitudeinthescanlinesafterfilteredand envelope-detectedaremappedtothebrightnessorgraylevelofthepixel. There are many kinds of commercialized probes used for medical diagnoses. The transducer can be dividedintosingleelementtransducersandarraytransducers. Currently,arraytransducersarepreferably used in ultrasound but in some specific cases, single element transducers are used. Figure 1.2 shows singleelementtransducersandanarraytransducer. Theprincipalfeaturesofsingleelementtransducers are self-focusing, mechanical scanning, simple fabrication, and adaptable for high frequency. On the contrary, array transducers use electrical scanning, multiple transmit focusing, receive focusing, low mechanical noise and high frame rate. Nowadays, three different types of 1D array transducer: linear array, carved array and phased array transducers are mainly used. Figure 1.3 shows the field of view of each array. As one of the examples, the utility form of transducers and frequency ranges for ultrasonic imagingisshowninTable1.1,whichiscourtesyofGEhealthcare(http://www3.gehealthcare.com/). 5 1.1 MedicalUltrasound Figure1.1: Ultrasoundprobes. CourtesyofGEHealthcare. Comparison of ultrasound to other modalities such as X-ray, CT, and MRI is tallied in Table 1.2. Ultrasound images are highly detailed and geometrically correct to the first order. These maps of the mechanical structures of the body depend on density and stiffness or elasticity, which determine the acoustic impedance of the structures. The dynamic motion of organs such as the heart can be observed by ultrasound. Diagnostic ultrasound is noninvasive, regarded as safe and does not accumulate any biologicalsideeffects. Additionally,itisrelativelylowcostandportability. However,ahighskilllevelis importantforobtainingsufficientlyevaluableimages,thatis,ultrasoundimagesareoperator-dependent. Ultrasound is not available to view bone, lung or any other gaseous parts because sound wave cannot penetratethembutreflectedattheseboundaries. Conventional X-Ray imaging is more straightforward than ultrasound which easily diffracts due to its long wavelength. As a matter of fact, because of the straight ray paths, accurate images are obtained inageometricsense. Despiteofshortexposuretime, X-raysareaformofionizingradiation,sodosage 6 1.1 MedicalUltrasound Table1.1: Anexampleoftransducerguide,courtesyofGEhealthcare Transducer Description Application Footprint Bandwidth lineararray Peripheral vascular, small parts, superficial muscu- loskeletal, nerve blocks, thoracic/pleural, needle guidanceandopthalmic 12.7x47.1mm 8.0-13.0MHz lineararray Peripheral vascular, small parts, nerve blocks, nee- dle guidance and superfi- cialmusculoskeletal 11.1x34.8mm 8.0-18.0MHz phasedarray Audlt and pediatric cardiac, abdomen, thoracic/pleural and opthalmic 19.3x27.6mm 1.5-4.0MHz convexarray Abdomen (including bladder) and conven- tional musculoskeletal (includinghipandspine) 18.3x66.2mm 2.5-6.0MHz microconvex array Obstetrics, gynecology andtissuebiopsy 26x4.3mm 3.48-9.0MHz microconvex array Abdominal, pediatric, small organ, neonatal cephalic, superficial musculoskeletal, tho- racic/pleural, intraopera- tiveandopthalmic 12x22mm 5.5-10MHz 7 1.1 MedicalUltrasound (a) (b) Figure 1.2: Ultrasonic Transducer, (a) single element transducer, courtesy of Olympus and (b) linear array,courtesyofIntelligenceofmedicaltechnologies. effectscanbecumulative. Computed tomography produces three-dimensional images and also involves X-rays. Based on the variation of attenuation of X-ray dependent on tissue density, the images are reconstructed by using uniqueprocessingscheme. Theapparatusofcomputedtomographyislargeenoughtofitapersoninside, resulting in being more expensive. The overall dose is much higher than that of a conventional X- ray. Computed tomography creates superb images of the brain, bone, lung and soft tissue, which can compensateforthecapabilityofultrasound. Because the body is highly water-content, magnetic resonance has been successfully developed and utilizedtoobtainimagesofsoftandhardtissuesinthebody. Thesubjectisplacedinastrongmagnetic field created by a large enclosing electromagnet that increases the cost and the need for the large space. The advantage is non-ionization and three dimensional images but the disadvantage is to take long time toformanimage. 8 1.1 MedicalUltrasound Figure 1.3: Field of view of 2-D images corresponding to (a) linear array, (b) convex array, and (c) phasedarray. High frequency ultrasound means ultrasound with higher frequency than conventionally used in medicine: higher than 20 MHz. In ophthalmology, dermatology and small animal imaging, scanners operated at the high frequencies called ultrasonic backscatter microscope or ultrasonic biomicroscope (UBM) have been developed, which obtain images by scanning a single element ultrasonic transducer in a sectorial format or linearly [1]. The resolution of ultrasound image is proportional to frequency so that it is useful to make high resolution images by high frequency ultrasound but the attenuation of ultrasound in the tissue is also approximately proportional to the frequency. As a result, the applica- tion of high frequency ultrasound is naturally limited: we cannot observe the fetal inside of its mother with high frequency ultrasound because of its penetration. Additionally, the electronics for imaging at high frequencies could be more complex and expensive. Figure1.4 is a picture showing the electronics compatible to 30 MHz and 128ch, experimentally developed in Ultrasonic Transducer Resource Center (UTRC)atUniversityofSouthernCalifornia(LosAngeles, CA). Acommercial systemnotformedical but for experimental research is shown in Figure 1.5. Furthermore, the fabrication and design of high frequency ultrasonic transducer array have the difficulty because it has finer pitch and requires good materialforthearraytobematchedtoelectricalimpedanceofelectricalcomponents. 9 1.1 MedicalUltrasound Table1.2: ComparisonofImagingModalities ImagingModality Ultrasound X-ray CT MRI Images Longitudinal, shear, mechanical properties Mean X-ray tis- sueabsorption Local tissue X-ray absorption Biochemistry Access Small windows ad- equate 2sidesneeded Circumferential aboundbody Circumferential aboundbody Spatialresolution Frequency and axi- ally dependent 0.2- 3mm 1mm 1mm 1mm Penetration Frequency depen- dent,3-25cm Excellent Excellent Excellent Safety Excellent for > 50 years Ionizing radia- tion Ionizingradiation Verygood Speed >100frame/sec minutes 20minutes 45 minutes (10 frames/sec) Cost Reasonable Good Expensive Expensive Portability Excellent Good Poor Poor VolumeCoverage Real-time 3D vol- umes,improving 2D Large3DVolume Large 3D Vol- ume Contrast Increasing(shear) Limited Limited Slightlyflexible Intervention Real-time 3D in- creasing No(Fluoroscopy) No Yes,limited The high frequency arrays being researched at UTRC and other institutes. Cannata et al. [13] re- searched 30 MHz, 256ch linear array fabrication by using 2-2 piezo composite. R. Chen et al. [14] reported a 40 MHz, 32ch phased array made of PMN-PT single crystal. More recently, Cummins et al. [15] attempted to develop 70 MHz miniature linear array to assist ultrasound-guided breast biopsy. The realization of high frequency ultrasound array will confront a number of issues. For instance, the electronics compatible to high frequency has to be developed. And, manufacturing these arrays at low cost will have to be achieved. In addition, the application of high frequency ultrasound array is limited sofar. Tocommercializehighfrequencyarrays,theseissuesmustbesolved. In the particular case of using a single element transducer, high frequency ultrasound has already 10 1.1 MedicalUltrasound Figure 1.4: Imaging platform developed in Ultrasonic Transducer Resource Center at University of SouthernCalifornia. Figure1.5: Imagingplatformforresearch,courtesyofVerasonics. 11 1.1 MedicalUltrasound been utilized in several fields. One of the most successful application in high frequency ultrasound is intravascular imaging which is the technology to image the wall of blood vessels in order for the esti- mation of the extent of stenosis and the characterization of atherosclerotic plaques by rotating a single elementhighfrequencyultrasonictransducer. Intravascularultrasoundscannersaretypicallyoperatedin thefrequencyrangefrom20MHzto60MHzandmountedonthetipofacatheter. Volcanoisaleading company in this field. Figure 1.6 shows an example of the intravascular ultrasound. What should be worthyofspecialmentionintermsofintravascularimagingisutilizationofultrasoundtissuecharacter- izationtechnologybasedonbackscatteringfeature. Integratedbackscatteringisuniquelyappliedforthe intravascularimagingforthepurposeofthedifferentiationofplaquetissueslocatedinthewallofvessel. Thecolorsshowninthefigure1.6indicatethenatureofplaquetissues. Figure1.6: Intravascularultrasound,courtesyofVolcano. Although not all of innovation in ultrasound is explained in the chapter, new attempts in ultrasound 12 1.2 BreastCancer such as the introduction of multi-modality imaging combining ultrasound with optical coherence to- mography to compensate for the resolution and the area especially in the vicinity of the surface [16], 4D ultrasonic imaging realizing real time observation of moving matter inside of bodies, and ultra- sonic tweezers based on ultrasound force to manipulate small particles like hemoglobin [17] have been conducted vigorously. Considering these technologies and trends mentioned in the paragraphs above, ultrasoundtechnologyisundoubtedlygoingtoprogresswithoutstagnation. 1.2 BreastCancer Breast cancer is an uncontrolled growth of breast cells as same as any other cancers. The term "breast cancer" refers to a malignant tumor that has developed from cells in the breast. Breast cancer usually beginseitherinthecellsofthelobules: themilk-producingglands,ortheducts,passagesthatdrainmilk from the lobules to the nipple. Cancerous cells by degrees can invade nearby healthy breast tissues and approach themselvesto the underarm lymph nodes. Then they attain the pathway into other parts of the bodycalledmetastasis. Breastcancerisalwayscausedbyageneticabnormalityinheritedfrompatient’s parentsorresultingfromagingprocess. Figure 1.7 shows the anatomy of breast. Breast consists of the adipose, glandular, and connective tissuesborderedbyanteriorandposteriorleafofsuperficialfasciaofthethorax. Adiposetissuescanbe observedasfattylobesthatareincorporatedintoglandulartissueandsurroundedwithconnectivetissue fibers. Glandular tissues of the breast are covered with a superficial layer subcutaneous fat. Terminal ducts lobular units are the basic functional units of the breast that produce milk. There are several generations of lactiferous ducts ending with subsegmentary and segmentary ducts and lactiferous sinus (Figure 1.8). Terminal ducts originating from every lobule run into ducts of the second generation, the latter running into the ducts of the third generation. Before draining to the nipple, the major ducts form lactiferous[18]. 13 1.2 BreastCancer Figure1.7: Breastanatomy Figure1.8: Lactiferousduct Breast cancer can begin in different areas of the breast: the ducts, the lobules, or in some cases, the tissueinbetween[20]. Ductalcarcinomainsitu Ductal carcinoma in situ (DCIS) is the most common type of non-invasive breast cancer. It has not spread beyond the milk duct into any normal surrounding breast tissue. Although DCIS does not threat the patient’s life, afterwards, the risk of development into an invasive breast cancer can increase. Thegrowthofductalcarcinomafromnon-invasivetoinvasiveisshowninfigure1.9. Invasiveductalcarcinoma Figure 1.10 shows the schematic of invasive ductal carcinoma (IDC), which is the most common typeofbreastcancer. About80percentsofallbreastcancersareIDC.IDCmeansthecancercells originated from ducts have broken the basement membrane surrounding the ducts and invaded another breast tissue around them. It is possible to approach to the lymph nodes, which can be metastasis. Aswomenbecomeolder,IDCismorecommon,thoughitaffectswomenatanyage. 14 1.2 BreastCancer Invasivelobularcarcinoma Invasive lobular carcinoma (ILC) is the second most common type of breast cancer. It refers to canceroriginatedfromlobuleshavingbrokenthroughthewallofthelobuleandinvadedthetissue ofthebreast. ILCcaneventuallyspreadtothelymphnodesandtootherareaofthebody. Aswell asIDC,ILCismorecommonaswomenbecomeolder. IttendstooccurlaterinlifethanIDC.Itis suggested that the use of hormone replacement therapy during and after menopause can increase theriskofILC. Lobularcarcinomainsitu Lobular carcinoma in situ (LCIS) is not regarded as cancer in spite of including the word: carci- nomabutoneoftheindicationoftheriskofbreastcancerinthefuture. LCISmeanstheexistence of abnormal cells in lobules without spreading to surrounding breast tissues. LCIS is usually di- agnosedbeforemenopauseandisuncommoninmen. MammogramcannotdetectLCISbutitcan be diagnosed as a result of a biopsy performed on the breast. This is the reason why LCIS is not wellknownhowmanypeopleareaffected. Inflammatorybreastcancer Inflammatory breast cancer is a rare and aggressive form of breast cancer so it is important to recognize symptoms and seek prompt treatment. Inflammatory breast cancer usually starts with thereddingandswellingofthebreastinsteadofadistinctlumpandhasatendencytoexpandand spreadquickly,withsymptomsworseningwithindaysevenhours. ThedefinitionofbreastcancerstagesispresentedonTable1.3. Basedontheresultofpathology,the stageofthebreastcancercanbereported. Thecancerstageisbasedonfourcharacteristics: 15 1.2 BreastCancer Table1.3: Stagesofbreastcancer Stage Definition Stage0 CancercellsremaininsidethebreastductsuchasDCIS,withoutinvasionintonormaladja- centbreasttissue. StageIA Thetumormeasuresupto2cmANDthecancerhasnotspreadoutsidethebreast-nolymph nodesareinvolved. StageIB There is no tumor in the breast; instead, small groups of cancer cells – larger than 0.2 millimeterbutnotlargerthan2millimeters–arefoundinthelymphnodes OR thereisatumorinthebreastthatisnolargerthan2centimeters,andtherearesmallgroups of cancer cells -larger than 0.2 millimeter but not larger than 2 millimeters - in the lymph nodes. StageIIA Notumorcanbefoundinthebreast, butcancercellsarefoundintheaxillarylymphnodes (thelymphnodesunderthearm) OR thetumormeasures2centimetersorsmallerandhasspreadtotheaxillarylymphnodes OR thetumorislargerthan2butnolargerthan5centimetersandhasnotspreadtotheaxillary lymphnodes. StageIIB The tumor is larger than 2 but no larger than 5 centimeters and has spread to the axillary lymphnodes OR thetumorislargerthan5centimetersbuthasnotspreadtotheaxillarylymphnodes. StageIIIA No tumor is found in the breast. Cancer is found in axillary lymph nodes that are sticking togetherortootherstructures,orcancermaybefoundinlymphnodesnearthebreastbone OR the tumor is any size. Cancer has spread to the axillary lymph nodes, which are sticking togetherortootherstructures,orcancermaybefoundinlymphnodesnearthebreastbone. StageIIIB Thetumormaybeanysizeandhasspreadtothechestwalland/orskinofthebreast AND may have spread to axillary lymph nodes that are clumped together or sticking to other structures,orcancermayhavespreadtolymphnodesnearthebreastbone. InflammatorybreastcancerisconsideredatleaststageIIIB. StageIIIC Theremayeitherbenosignofcancerinthebreastoratumormaybeanysizeandmayhave spreadtothechestwalland/ortheskinofthebreast AND thecancerhasspreadtolymphnodeseitheraboveorbelowthecollarbone AND thecancermayhavespreadtoaxillarylymphnodesortolymphnodesnearthebreastbone. StageIV Thecancerhasspread-ormetastasized-tootherpartsofthebody. 16 1.2 BreastCancer Figure 1.9: Ductal carcinoma in Situ, show- inghowductalcarcinomainsitudevelopsfrom non-invasivetoinvasivecarcinoma[20]. Figure 1.10: Invasive ductal carcinoma. A:Ducts, B: Lobules, C: Dilated section of duct to hold milk, D: Nipple, E:Fat, F: Pec- toralis major muscle G: Chest wall/rib cage. In enlargement, A; Normal duct cells, B: Duc- tal cancer cells breaking through the basement membrane,C:Basementmembrane[20]. thesizeofthecancer whetherthecancerisinvasiveornon-invasive whethercancerisinthelymphnodes whetherthecancerhasspreadtootherpartsofthebodybeyondthebreast Each stage is correspondent with 5 year survival rate. If the tumor is diagnosed as Stage I breast cancer,thepatient’s5yearsurvivalrateis98%butifpatientisdiagnosedofStageIV,herprobabilityof survivalin5yearsislessthan16%. 17 1.2 BreastCancer Table 1.4: Breast cancer statistics in the U.S. and Japan, the numbers are population of new cases and deathinayearandthenumbersinsideofparenthesesaretheratesper100,000ofeach. UnitedStates a Japan b invasivebreastcancer(rate) 224,147(122.2) 72,472(110.5) non-invasivebreastcancer(rate) 60,290(30.4) 8,847(5.2) deathcausedbybreastcancer(rate) 41,150(21.3) 13,148(20.4) c The statistics on breast cancer in the U.S. and Japan are shown in table 1.4. In the U.S., the rates of newcasesanddeathtendtodecreasebecauseofthetreatmentadvances,earlierdetectionthroughscreen- ing,andincreasedawarenessofbreastcancers. Incontrast,theratesoftheminJapanareincreasing. One of the reasons is that the rate of women receiving routine inspection such as mammography and ultra- soundinJapanismuchsmallerthanintheU.S.becausethenumberofpeoplewhodiedcausedofbreast cancer used to be small. However, nowadays, the fooding culture in Japan was changed or introduced intoEuropeanstyle. Otherreasonthatmaybeincommonintheworldisthatalifetimebecomeslonger than that used to be. The most significant risk factors for breast cancer are gender and age according to Breastcancer.org[20]. The death rate of breast cancer is higher than that of any other types of cancer in the U.S., except lung cancer. Its rates in Japan are the fifth highest, and the first is that of large intestine cancer. Besides skincancer,breastcanceristhemostcommonlydiagnosedcanceramongAmericanwomen. Concretely describing,under30percentsofnewlydiagnosedcancersinwomenwillbebreastcancer. Interestingly, the rate of Japanese women who were newly diagnosed of breast cancers to all new patients of cancers was the highest of the rate of those who became all other cancers in Japanese women. The reason of thedifferenceofthetendencybetweenAmericanandJapaneseisthatthenumberofskincancerpatients is strongly dependent on races: white people become skin cancer more easily than any other people. a 2012[21],[22] b 2011[23] c 2013[23] 18 1.2 BreastCancer In describing the dependency on races of breast cancer, white women are a little more likely to develop breastcancerthanAfrican-Americanwomenbutinyoungerwomen,African-Americanwomenaremore common than white, and the death rate of breast cancer is higher than that of white women. The risk of developingbreastcancerislowerinAsian,HispanicandNative-Americanwomen[21]. It is undoubted that external factors are primally responsible to the development of most types of cancer including breast cancer. However the development of some cancers seems to be intensively re- latedtohereditybasedontheresearchtodate[24],[25]. Thankstotherecentsteepinnovationofgenetic engineering and the development of analytical tools of genetic information, numerous numbers of rela- tionship between genetic information and diseases have been gradually disentangled in this 20 years. It had been suspected and is indicated with certainty that there are some genes which can be susceptive to breast cancer by spontaneous mutation. About from 5 to 10 percents of breast cancers can be linked to gene mutations inherited from patient’s mother or father. The mutation of the BRCA1 and BRCA2 genes are the most common. Women with a BRCA1 mutation have about 50 percents of lifetime risks and the risk of those with a BRCA2 mutation is 45 percents. About 85 percents of breast cancers occur inwomenwithoutfamilyhistoryofbreastcancer,whichresultsfromgeneticmutationscausedbyaging process and life. It should be added that these mutations are relevant to prostate cancer that are in men sothegenemutationcanbefromtheprospectivepatient’sfathersaswellastheirmothers[18]. The diagnosis of breast cancer including the addition of genetic inspection has been improved re- cent decades. The diagnosis of breast cancer consists of genetic inspection, ultrasound, mammography, magnetic resonance imaging mammography and some biopsies. Usually, genetic inspection is not used because it is not covered with typical insurance plans and it is expensive in U.S. Recently in Japan, however, genetic inspection started to be offered at a reasonable price, though it is not covered by any general insurance. So it can be a conventional scheme of screening cancers and other diseases. Follow- ings will introduce some methods of the diagnosis: genetic inspection, ultrasound, mammography and ultrasound-guidedbreastbiopsy. 19 1.2 BreastCancer Geneticinspection Genetic testing is necessary to progress step by step deliberately because the genetic information is exactly personal information which is a risk of security. At first, counseling is conducted to collectpaternalandmaternallineages. Ifthegenetictestingisappropriate,whoreceivethetestis determined. Genetic testing can be done in a research laboratory as well as a clinical laboratory, but the techniques vary so it is important which laboratory conducts the testing. The time for the genetic testing depends on the technique of the lab and on the patient’s genes. BRCA1/2 genetic testing results are typically available within two weeks of the blood draw. The result can affect surgical decision-making if the result are available before definitive surgery. After identification of the lab, informed consent is must. Once the result of the genetic testing is available, it should bedisclosedtothepatientimmediately[18]. | Advantage: avoidableover-screeninggoodreferenceofscreeningandoperatingplan | Disadvantage: expensive still skill dependent security concerns less adaptation of most of insuranceplans Ultrasound Although mammography is still gold standard for breast cancer screening, breast ultrasound has also been predominantly used as a target examination for a clinical and mammographic problem. InEuropeandJapan,whole-breastultrasoundsurveyhasbeenmoreprevalent,whichisadifferent from in U.S. It is not unusual to identify cancers which cannot be found by touch, diagnosing by ultrasound. It is known that mammography is not perfect and have a limitation of sensitivity, especially for women whose breasts contain dense tissue. The breast ultrasound is regarded as supplemental imaging modality for the screening in women with dense tissue and in those with high risk for breast cancer. Breast ultrasound is performed with linear probe with the frequency of 5 - 12 MHz, more often 7.5 - 10 MHz. Figure 1.11 shows the situation of breast ultrasound 20 1.2 BreastCancer examinationandfigure1.12presentsthepictureofnormalbreastultrasound. | Advantage: Identifiable of small node-negative cancers that can be missed by screening mammographybettertoleratedbythepatient,notionizingradiation,nopatientdiscomfort | Disadvantage: Operator dependent requirement of long physician time high false-positive ratelowsensitivityforDCIS Figure1.11: Positionofultrasoundprobeinbreastultrasound[19]. Mammography Mammography is the examination of the human breast using low-energy X-rays, which deserves the best available and the only imaging modality scientifically proven to be worthy of reducing mortality. The essential goal of mammography in screening is to find high rate of small cancers whichcannotbedetectedbytouch(nonpalpable),keepingthefalsepositiveacceptablylowlevel. Figure 1.13 shows the image of mammography examination. Figure 1.14 presents the picture of mammographyforscreeningbreastdiseases. 21 1.2 BreastCancer Figure1.12: Normalbreastultrasoundandgraphic[19]. 22 1.2 BreastCancer | Advantage: detection of impalpable breast lesions high diagnostic value possibility of inva- siveandnoninvasivediagnosticproceduresobjectivedocumenteddataccessiblefordynamic analysis | Disadvantage: IonizingradiationLowvalueindenseandirregularstructureofthebreast Figure 1.13: Mammography, courtesy of UniversityofConnecticutHealthCenter. Figure 1.14: Standard images obtained forscreeningmammography[19]. Ultrasound-guidedbreastbiopsy Ultrasound-guided breast biopsy which is one method of the breast intervention that has evolved over recent 20 years. The rate of open surgical biopsy has dramatically dropped during this pe- riod because the technique identifies the position of abnormality which should be detected and diagnosed in pathology. Especially ultrasound is advantageous due to non-ionizing radiation, cost-effectiveness, real time imaging, accessibility, non-uncomfortableness and quickness. Core needle which has a larger diameter than a simple fine needle is frequently used in breast biopsy. Ultrasound-guidedcoreneedlebiopsyisusedtoevaluatesuspiciousandhighlysuspiciouslesions thataredetectedbyultrasoundforsurgicalplanningorneoadjuvanttherapy. Anditisalsousedfor probably benign lesions to avoid more expensive and invasive surgical biopsy. Figure 1.15 shows ultrasound-guided fine needle aspiration biopsy and Figure 1.16 shows vacuum-assisted biopsy 23 1.2 BreastCancer whichisoneofthecoreneedlebiopsy. Lineararraywithfrequencyofnotlessthan10MHz,hope- fully 12.5MHz should be used to perform optimal ultrasound-guided core needle biopsy. Like figure 1.17, the needle must be parallel to the azimuthal axis of the linear array and be as parallel tothechestwallaspossible. Figure1.15: Fineneedleaspirationbiopsyofthebreast[19]. As mentioned so far, ultrasound-guided breast biopsy is a pivotal clinical method for follow-up ex- amination and surgical planning. Despite of the high diagnostic accuracy of needle biopsy and the fact that over 1.6 million women undergo breast biopsy annually in the United States, thousands of patients receivefalsenegativediagnosis,whichcancausecancerprogressionandincreasingmortality[26],[27]. The missed diagnoses attribute to the inaccuracy of tissue sampling at the time of biopsy [28]. Micro- calcification which may suggest early breast carcinoma such as DCIS and LCIS cannot be detected by usingconventionalultrasoundwhosefrequencyrangefrom2MHzto20MHzbecauseofpoorresolution 24 1.2 BreastCancer Figure1.16: Vacuum-assistedbiopsyofthebreast[19]. Figure1.17: Prefirelocationandpostfirelocationofultrasound-guidedneedlebiopsy[18]. that is proportional to ultrasound frequency. Transcutaneous ultrasound with high frequency attenuates and does not receive high contrast echo signal even if the coefficient of reflection at the surface of mi- crocalcification is much higher than any other tissues. Ultrasonic Transducer Resource Center having researchedhighfrequencyultrasoundsuggestedanddevelopedahighfrequencyultrasoundarraywhich 25 1.3 ObjectiveoftheResearch wasdesignedtobeintegratedintheneedleforbreastbiopsy. Thedevicecanbesignificantforultrasound- guidedbiopsy becausethe radiologists could watchultrasound images not only out of skin but from the biopsy needle with the high frequency ultrasound array that can generate high resolution images which can resolve microcalcifications. Therefore, they can locate lesions and obtain tissue samples in breast moreaccuratelythanwithoutusingtheultrasound-integratedbiopsyneedle[15]. 1.3 ObjectiveoftheResearch The goal of this research is to develop the novel breast biopsy needle with the capability of the accu- rate guidance for taking the tissue that physicians aim to acquire for diagnostic pathology. Figure 1.18 showsapictureofthebreastbiopsyneedlewithaminiatureultrasonictransducerarray,thoughthisisan experimental piece. The proposed ultrasound-guided breast biopsy utilizes the percutaneous ultrasound guidance with the low frequency ultrasound combined with a miniature high frequency ultrasound de- vice. The concept is shown in figure 1.19. The needle is guided by the percutaneous ultrasound and it can detect small structures such as microcalcifications that cannot be observed by using low frequency ultrasound. Thecombinationoflowandhighfrequencyultrasoundcanbeexpectedtoimprovelocaliza- tionofabnormaltissuesandtheaccuracyofbreastbiopsy,whichcanreducethenumberofpatientswho receivefalsenegativediagnoses;itcanbeasignificantinnovationforthediagnosisofbreastcancerthat is one of the most serious diseases in the world. To enhance the capability of the needle, in this thesis, thepotentialofhighfrequencyultrasoniccharacterizationofbreastbiopsytissueswillbedemonstrated. This can be one of the useful applications of high frequency ultrasound. In addition, building the test systemfortheevaluationofhighfrequencyultrasonictransducerarrayswillbeshown. Thissystemcan beavaluabletoolfortheresearchanddevelopmentofultrahighfrequencyultrasonictransducerswhich arestillincompatibletotheconventionaland/orcommercializedelectronics. 26 1.3 ObjectiveoftheResearch Figure1.18: Atrialpieceofabreastbiopsyneedlewithaminiaturehighfrequencyultrasonictransducer array[15]. Figure 1.19: The proposed concept of ultrasound-guided breast biopsy combining low and high fre- quencyultrasound. 27 Chapter2 BreastBiopsyTissue 28 2.1 BreastBiopsyTissue 2.1 BreastBiopsyTissue As described in the introduction of this thesis, breast biopsy is a gold standard in diagnosis of breast diseases. The biopsy using core needle is most common guided by ultrasound or mammography called stereotactic biopsy, which is often applied in case the lesion that should be acquired is located where ultrasound cannot detect or the size of lesion is too small to be detected by ultrasound. One of the prin- cipalreasonswhythebreastbiopsyissignificantisbecausethisisonlyonemethodthatcancharacterize the tissue in breast and identify whether the lesion in breast is benign or malignant, whether it is still in the specific area or possible to spread around it, which mammography and conventional ultrasound do not currently have the ability to do so. In this chapter, It will be described what types of cells are observedinbreastbiopsytissue,theprocedureofpathologyonbreastbiopsytissueisexecutedandwhat ishistologicalimagesofbreasttissueinordertopresentmyfocusofthisresearch. 2.1.1 Breastcellsandtissues Asdescribedinsection1.2,breastconsistsoftheadipose,glandularandconnectivetissuesborderedby anteriorandposteriorleavesofsuperficialfasciaofthethorax. Somekindsoftissueswhichareobserved inbreastbiopsyaregoingtobedescribedbelow[19]. Adiposetissue Adiposetissuecanbeobservedasfattylobeswhichareincorporatedintoglandulartissueandsur- rounded with connective tissue fibers. Adipose tissue known as subcutaneous fat is characterized with decreased or normal echodensity and homogenouse-enough structure with linear echogenic incorporations, which often exhibit vague acoustic shadows. The thickness of adipose layer de- pends on the patient’s constitution but as rule, increases with age. In young women, the adipose tissueispresentedbyathinlayerbetweentheskinandglandulartissuebutwithage,afterpregnan- ciesandchildbirths,thethicknessofthisadiposelayerincreasesaccompaniedwithalittleincrease 29 2.1 BreastBiopsyTissue in echodensity. In postmenopause with the beginning of mammary involution, the adipose tissue becomesmoreirregularduetothedevelopmentofconnectivetissue. Glandulartissue Glandular tissue of the breast is covered with a superficial layer subcutaneous fat. Its thickness depends on the age and constitution of the woman. Glandular tissue so called parenchyma which consistsofalveolar-tubularcomplexesnormallylookslikealayerwithslightlydecreasedechoden- sity and irregular echostructure. The echodensity of glandular tissue can vary from decreased to increased, dependent on the age. The echodensity of parenchyma can also change depending on thephaseofmenstrualcycle. Proliferativeprocessesdecreasetheechodensityofglandulartissue. In young women, glandular tissue is usually isoechoic or slightly hyperechoic and is observed as arelativelyhomogeneousuniformlayer. Connectivetissue Connective tissue is included into the structure of fibrillar tissue between glandular elements and in the walls of lactiferous ducts. It produces septa called Cooper’s ligaments, thus forming the breast "skeleton" and maintaining structural integrity. The Cooper’s ligaments become thicker andoftenfromadiposelobuleswithlateralacousticshadows. Connectivetissuefibershavesmall amount of collagen and are practically invisible against the basic layer. Only in the second phase of menstrual cycle there can arise granularity due to echogenic fields of glandular tissue against hypoechoiclactiferousducts. So far, several normal breast tissues were described. The typical ultrasound image of breast has already shown in figure 1.12. In breast biopsy tissues, however, abnormal tissues are probably con- tained and imaging them are significant for diagnosis of breast diseases. Numerous types of abnormal tissuesandcellsexistbutsomeofthemwhicharerepresentativeandwerefoundinthisresearchwillbe describedbelow. 30 2.1 BreastBiopsyTissue Microcalcification Microcalcifications are small deposits of calcium in the breast. Diffrent from macrocalcifications that are found in around half of all women over the age of 50 and that may be caused by cal- ciumdepositsinmilkducts,previousinjuriesorinflammation,microcalcificationscansometimes be a sign of precancerous changes or early breast cancer if a group of them is found one spe- cific area though they are not usualluy due to cancer. Microcalcification are usually observed by mammogram and are sometimes viewed by ultrasound. Figure 2.1 shows a histological image of microcalcificationsandfigure2.2showsmammogramofthemthatareclusteredwithpartlyductal orientationwhichcanindicatetheyprobablyresultfromsortoftumorsuchasDCIS[18]. Figure 2.1: Microcalcifications. Artificial dis- ruptionoftissue[30]. Figure 2.2: Mammogram of microcalcifica- tions. [18]. Adenocarcinoma Adenocarcinoma like figure 2.3 is a type of carcinoma which is a cancer that begins in the lining layer of organs such as breast, lung, pancreas and so on, specifically starts in gland tissue. The ductsandloublesofthebreastareglandtissuesbecausetheymakebreastmilk,socancersstarting in these areas are often called adenocarcinomas. Greater than 95% of breast malignancies are adenocarcinomas, which are devided into in situ carcinomas and invasive carcinoma. Carcinoma 31 2.1 BreastBiopsyTissue in situ refers to a neoplastic proliferation that is limited to ducts and lobules by the basement membrane. On the other hand, invasive carcinoma has penetrated through the membrane into stroma[29]. Figure2.3: Adenocarcinoma,10timesmagnification. [31] In this study, the breast biopsy tissue specimens consist of the tissue types that were introduced previously but the breast tissue can contain much more kinds of cells which are abnormal. However it is impossible to obtain specific abnormal cells from the patients because nobody knows which kinds of cellspatientshaveuntilpathologyisconducted. Thisstudyattainedtodealwithrelativelytypicaltissues andcellsbyhighfrequencyultrasoundandpathology. 2.1.2 Methodofpathology The procedure of pathology is composed of biopsy, fixation, embedding, section cutting, mounting on glass slides, rehydration, staining, mounting medium and diagnosis. In this section, these steps will be introduced orderly, which are basically general steps besides diagnosis but in this study, ultrasonic imagingwasintervenedbeforediagnosticpathology,thensomeirregularstepswereappliedandwillbe alsoshowninthissection. 32 2.1 BreastBiopsyTissue Biopsy Biopsy is removal of a sample of living tissue for microscopic examination. Specifically, in case withbreastbiopsy,fineneedle,coreneedle,vacuumassistedcoreneedleandsurgicalexcisionare usedfortheremoval. Coreneedleswithguidanceofultrasoundandmammographyarefrequently applied to acquire the target tissue accurately. Figure 2.4 shows the needles and tissues that are removed by using core needles. For example, 14 gauge needle has nominal outer diameter of 2.1mm and 16 gauge needle has that of 1.7mm. In vacuum assisted core needle, 14 gauge can obtainthevolumeoftissueof34to40mg[18]. Afterbiopsy,biopsymarkercanbeplaced,which is essential to specify the place at which tissue are removed, that is optimal patient management. Themarkercanbedetectedbymammographyafterbiopsy. (a) (b) Figure 2.4: Core needle breast biopsy [18]. (a) Automated spring-loaded core needle tip in the unfired (top)andfired(bottom)position. (b)Sizeofa16-gaugesamplesversus14-gaugesamplesobtainedwith anautomatedspring-loadedcoredevice. Fixation Fixationisacomplexseriesofchemicalevents. Theaimofthefixationistopreventautolysisand bacterialattack,tofixthetissuessotheywillnotchangetheirvolumeandshapeduringprocessing, topreparetissueandleaveitinaconditionwhichallowclearstainingofsectionsandtoleavetissue 33 2.1 BreastBiopsyTissue as close as their living state as possible so that no small molecules should be lost. The fixation is coming by reaction between the fixative and protein which form a gel, so keeping every thing as theirinvivorelationtoeachother. Thesampleobtainedbybiopsyaredippedwithfixativesuchas 10% formaldehyde immediately. 10% formaldehyde is mostly used but in this study Dulbecco’s phosphatebufferedsaline(LonzaGroup,Ltd.,Basel,Switzerland)wasused. Embedding Theaimofembeddingistomakeablockwithtissueinasolidmediumfirmenoughtoenablethin sectiontobecut,andsoftenoughtopreventfromdamagingtheknifeortissue. Afterdehydration and clearing step which are not explained in detail here, the embedding is the process by which tissues are surrounded by a medium such as agar, gelatin, or paraffin wax which will provide sufficient external support by solidifying during sectioning. In this study, a 3% agar gel (Fisher Scientific,Waltham,MA)wasusedbecauseofthesimplicityandlowattenuationofultrasound. Sectioncutting Cutting is by using the microtome (figure 2.5) which is a mechanical instrument used to cut bio- logical specimens into very thin segments for microscopic examination. Most microtomes use a steelbladeandareusedtoprepare sectionsof animalor planttissues forhistology. Slicedtissues aremountedonpiecesofslideglass. Thethicknessofthetissuesarelessthan5μm. Staining Hematoxylin and Eosin (H&E) are universally used for routine histological examination of tissue section. They stains the tissue sufficiently dried and mounted on the slide glass. Hematoxylin stains acidic molecules shades of blue and eosin stains basic materials shades of red, pink and orange. The results of H&E indicate nuclei of blue with some metachromasia and cytoplasm showsvariousshadesofpinkwhichidentifiesdifferenttissuecomponents. 34 2.1 BreastBiopsyTissue Figure2.5: Sectionsbeingcutfromaparaffinblockusingarotarymicrotome. CourtesyofLeica. After staining, the tissue on the slide glass is covered by a piece of cover glass, that is completion as a specimenforadiagnosticpathology. 2.1.3 HistologicalImageofBreastTissue Figure 2.6 shows slides for pathology. In most of cases on diagnostic pathology, optical microscope are used but scanning electronic microscopy (SEM) or transmission electron microscopy (TEM) are sometimesusedtocharacterizecells. Typicalhistologicalimagesofbreastdiseaseswillbeshownbelow. Figure2.6: Pathologymicroscopeslides. 35 2.1 BreastBiopsyTissue Figure 2.7: Normal breast, glands and stroma. Courtesy of Dr. Shashidhar in James Cook University,Australia. Figure 2.8: Fibroadenoma that is benign. In- tracanaliculargrowth. [30]. Figure2.9: Ductalcarcinomainsitu(DCIS)of cribriform pattern. Multiple spaces within the proliferation of monotonous cells are rounded anddistributedinanorganizedfashion[29]. Figure 2.10: Invasive ductal carcinoma, well differentiated(gradeI)[29]. 36 Chapter3 UltrasonicTransducerandImaging 37 3.1 UltrasonicTransducer 3.1 UltrasonicTransducer Oneofthemostindispensablepartsinultrasoundisanultrasonictransducerwhichisthedevicecapable of transmitting and receiving ultrasound, converting it into electrical signal and vice versa. The char- acteristics of the ultrasonic transducer determine the quality of ultrasound images, making it extremely essential. Although some ultrasonic transducers convert between electrical and acoustic or mechanical energy based on electrostatic principle, most use piezoelectric effect that can be observed in a material where applied mechanical stress results in an induced electric charge separation called direct piezo- electric effect. In contrast, an applied electric charge induces mechanical strain, which is called the reverse piezoelectric effect. An ultrasonic transducer based on the piezoelectric effect consists not only ofpiezoelectricmaterialthathasanabilitytochangeenergybetweenmechanicalandelectricalbutalso of electrodes that are situated both sides of the piezoelectric material. The transducer emits an acoustic wavewhenahighvoltagebetweenthetwoelectrodeisapplied. Thewavepropagatesthroughthepiezo- electricmaterialincludingreflectionsinsideofitandtoamedium. Thewaveresultsfrominstantaneous strainofthepiezoelectricmaterial. Apiezoelectricmaterialcandeformnotonlyinthenormaldirection about electrodes that are situated on opposite sides of the material but in other directions. Usually, the piezoelectricmaterialinadvanceispolarizedtodesirabledirection,inthiscase,normaldirectionagainst theelectrode. Inotherwords,mainstrainoccursbutundesirablestrainsdosimultaneously. Anyway,the pressure waves are generated from one of the surface of piezoelectric material. On the contrary, when a pressure wave comes from scatterers in the medium to an ultrasonic transducer, the piezoelectric ma- terial in the transducer produces a displacement and charging can be induced by the pressure wave, and thenelectricalsignalgeneratesfrombothelectrodesonthesurfaceofthepiezoelectricmaterialanditis recordedbytheimagingsystemandthroughanamplifierandfiltersinordertoenhancethecontrastand reduce noise of the image respectively. In this study, a single element ultrasonic transducer was used to verify the feasibility of the effectiveness of high frequency ultrasonic imaging for the improvement of 38 3.1 UltrasonicTransducer localization of abnormal tissues in performing the proposed ultrasound-guided breast biopsy that could beachievedbyusingaminiaturelineararraymountedonabiopsyneedle. Theirbasicstructureswillbe showninfollowingsections. Thefigure3.1showseachfundamentalelementofpiezoelectrictransducer inasingleelementtransducerandalineararray. Figure3.1: Piezoelectrictransducerelementgeometryfor(a)asingleelementandfor(b)alineararray. 3.1.1 SingleElementUltrasonicTransducer A single element ultrasonic transducer as shown figure 3.2 is the simplest form in all types of ultra- sonic transducers. This is a photo of a handmade single element transducer. And figure 3.3 shows the schematics of the section of single element transducers. A single element ultrasonic transducer is used for a mechanical scanning probe to obtain high resolution images by using high frequency ultrasound 39 3.1 UltrasonicTransducer applied on animals, skin, eyes and IVUS [1]. The minimum components of a single element ultrasonic transducer are piezoelectric material, electrodes, matching layer, backing layer for the reduction of un- desirable reflection and acoustic lens for focusing. Most piezoelectric materials for a single element ultrasonictransducerarethincylindricalshapelike(a)infigure3.1. Thethicknessofthemandmatching layer are designed for a target central frequency. What kinds of material should be selected is designed by using equivalent circuit model called Mason model [32] or using KLM (Krimholtz, Leadom, and Mettaei) model [33]. The diameter of the transducer is determined from the target center frequency of ultrasoundbutthetransducerhastobeelectricallymatchedtotheinputimpedanceoftheimagingsystem thatshouldbe50ohm,thenthepiezoelectricmaterialhastobeselectedunderthiscondition. Therefore, thefirststeptodesignthetransduceristoselectthepiezoelectricmaterial. Apiezoelectricmaterialispolisheddowntothetargetthicknessandelectrodesaredepositedonthe bothsideofthedieandturnedintocircularshape. Thebackinglayershouldbeacousticallymatchedto reduce reflection and to absorb the remaining wave. The material suitable for the backing layer is vis- cous resin with high impedance particle to adjust its acoustic impedance. For high frequency ultrasonic transducers, duetotheattenuationproportional tofrequency, thetip oftransducerhas aparabolic shape tofocustheacousticwaveinsteadofacousticlenswhicheasilyattenuateshighfrequencyultrasound. The methodology to produce ultrasound images is to spatially scan the probe to acquire echo signal ateachposition. Iftheprobeisscannedlinearly,theimageontherectangularplaneofaxialdirectionand scanning direction is produced. If the probe is scanned roundly, the round image around the probe can be obtained. One of the application of linear scan using a single element transducer is called ultrasonic biomicroscopy(UBM)thatwillbedescribedindetail. Theroundscannedimageiscommerciallyapplied for intravascular ultrasound called IVUS that can depict the images of the section of relatively large arterial and venous vessels via catheter. The purpose of the IVUS is the diagnosis and the operation of theirocclusionandsclerosisofthearteries(showninfigure1.6). 40 3.1 UltrasonicTransducer Figure 3.2: A photograph of single ele- mentultrasonictransducer. Figure 3.3: Schematics of lens-focused and press- focusedtransducers. UltrasonicBiomicroscopy(UBM) Ultrasonicbiomicroscopyaswellascalledultrasonicbackscattermicroscopyworkslikelytocon- ventional ultrasound but high frequency of between 20 and 50 MHz that rewards high resolution images. TheUBMexaminationtechniqueissimilartoB-scanbutdifferentfromitonthepointof scanning: movingatransducer. Asingleelementultrasonictransducerhasasinglefocuspointso that the image obtained by linearly scanning the transducer has only one axial distance that is the most highly resolved. Sometimes, the transducer is scanned to axial direction called B-D mode scan then the image has more resolved points than a standard linear scan of UBM at the cost of theframerate. TheUBMcanobtainimagesinlivingeyeswithoutaffectingtheinternalstructure and proves valuable in clinical practice. Figure 3.4 shows UBM for ophthalmology commercial- ized by Quantel Medical (Cournon-d’Auvergne, France). The UBM is also used for diagnosis in dermatologytoassesscutaneousdiseasessuchasmelanoma. Figure3.5isUBMfordermatology inDermascan(Hadslund,Denmark). Besidesclinicalapplication,theUBMisappliedtoresearch suchasin-vitrotissueimagesandsmallanimals. 41 3.1 UltrasonicTransducer Figure3.4: UBMforopthalmology. CourtesyofQuantelMedical,France. Figure3.5: UBMfordermatology. CourtesyofDermascan,Denmark. 3.1.2 LinearArrayUltrasonicTransducer A linear array ultrasonic transducer is one of the most straightforward array transducers that comprises multiple elements. The elements in a linear array are arranged in a line and are rectangular in shape. The array is sometimes called one-dimensional array for the distinction from other types of linear array such as 1.5D array and 2D array. The elements are located periodically with a distance called "kerf" thatisforsuppressionofcrosstalkbetweenelements. Eachelementisshapedlikefigure3.1(b). Dueto the limitation of geometry, the shape is typically rectangular. Therefore, even if the same piezoelectric materialisusedasadisc-shapedoneinasingleelementtransducer,thepiezoelectricfeaturesshouldbe differentbecausethepiezoelectricfeatureisdependentonthegeometricalshapeofthematerial. Figure 42 3.1 UltrasonicTransducer Figure3.6: Constructionofaone-dimensionalarraywithanelevationplanelens. [34] 3.6 shows the construction of a one-dimensional array that is one of the typical structure of 1D linear array,thoughSaitohetal. [34]introduceditofthephasedarraystructuretheystudied. The basic components of the array transducer is as same as a single element ultrasonic transducer: piezoelectricmaterial,twoelectrodessituatedontheoppositesidesofthepiezoelectricmaterial,match- ing layers, backing material and acoustic lens that focuses the imaging plane in the elevation direction andcontroltheslicethicknessoftheplane. Oneofthedifferencesiselectricallinesduetomultiplechan- nelsthathavetobeseparatedforpreventingfromelectricalshortandcrosstalk. Relatedtothis,backing materialmustbeofinsulationnotbeconductivelikeasingleelementtransducer. Anotherdifferentthing isaspacebetweenelementscalledkerfthatisforsuppressionofacousticormechanicalcrosstalk. Alineararrayultrasonictransducerisdesignedbyusingnotonlyequivalentelectricalcircuitmodel such as Mason and KLM model but high order simulation such as finite element method. One of the commercial tools is PZFlex (Weidlinger Associates, Inc. Mountain View, CA) that can calculate piezo- electric behavior, response on input signal, wave propagation through medium in 2 or 3 dimensions. 43 3.1 UltrasonicTransducer Figure3.7: Ultrasoundarraytechnologyprogressionwithdice-and-fill,lasermicromachiningandDRIE micromachiningfabricationmethods. [36] Thisismainlyusedforthepredictionofcrosstalkbetweenelements;thenanelementsizeandpitch,kerf widthanddepth,andotherstructuresofthearrayareabletobedesigned. The fabrication of ultrasonic transducer arrays is similar to that of a single element transducer but differentfromitregardingtheseparationofelements. Acompositeofpiezoelectricmaterialandpolymer filled between the piezoelectric materials that are machined into a plural of elements beforehand is pol- ished toward the targetthickness, which is applied for the fabrication of standard transducer arrays. For higher frequency array, the separation of element is executed by using laser dicing or deep reactive ion etchingtechnology. Thetechnologyprogressionofthefabricationofarraycompositeisshowninfigure 3.7. Ritter et al. [35] demonstrated the method of fabrication on 30 MHz linear array that composed of 2-2 piezoceremics composite, two matching layers, interconnects for signal lines, backing material and 44 3.1 UltrasonicTransducer Figure 3.8: Structure of 30MHz compos- itearray[35]. Figure 3.9: Photograph of 30MHz linear arrayassembly[13]. concave lens, which is shown in figure 3.8. Cannata et al. [13] demonstrated another fabrication tech- niquecalledinterdigitalbondingforhighfrequencylineararraytransducerwithmorechannelsandfiner pitch than before. The photograph of the array assembly is shown in figure 3.9. The probes clinically usedarealreadyshowninfigure1.1,figure1.2(b)andtable1.1. The scheme to produce ultrasound images with a linear array is to operate it by applying voltage pulses to a group of elements in succession. The pulses have some delays to generate a beam focused at target depth. The aperture determined which elements are excited is stepped by an element to define anextgroupofelementsliketheoneshownonfigure3.10(right)whereasamechanicalscannedsingle transducer is shown in the left of figure 3.10. The steps progress along to azimuthal direction and ultra- soundbeamisscanned,thatiscalledelectronicscanincontrastwithmechanicalscanonasingleelement transducer. A linear array ultrasonic transducer can scan acoustic waves without physical movement of itself then mechanical noise caused by the movement of the transducer does not occur. And a single element transducer has only one focal length while linear array can define multiple focal distances by changingthedelayofinputvoltagepulseandthenumberofelementsinagroupofaperture. 45 3.1 UltrasonicTransducer Figure3.10: Singleelementmechanicalsectorscannerandlineararrayscanner. [1] 46 3.2 UltrasonicImaging 3.2 UltrasonicImaging For medical ultrasound, gray scale imaging indicating B-mode, A-mode, and M-mode, Doppler flow measurement, harmonic imaging, contrast imaging are mainly used. In this study, quantitative ultra- soundthatisbackscatteringanalysistriedtobeappliedfortissuecharacterization. Inbiology,medicine andbiomedicalengineeringfield,statisticalanalysisissometimessignificanttovalidateclinicalandex- perimentalresults. Forthepurposeofitinthisstudy,somestatisticaltechniquesareapplied,whichwill beshowninthissection,thoughthetechniquesmaynotstronglyrelatedtoultrasonicimaging. Sincethis studyisforthedevelopmentofhigherfrequencyarraythanconventionalultrasoundthatiscompatibleto commercialimagingsystems,thesystemforevaluationofthearraymustbebuiltup. Syntheticaperture imagingtechniqueisappliedforit. 3.2.1 B-mode B of B-mode stands for brightness, that is to say, B-mode or B-scan indicates the image presenting brightnessmapping. TheB-modeisthemostprevalentinmedicalultrasoundandthemostfundamental in ultrasonic imaging. The voltage pulse generated by a pulse generator inputs to the transducer and it generates sound beam. In case with a single element transducer, the beam is naturally formed, whereas incaseanarraytransducerisused,agroupofelementstransmitsacousticwavesthatarefocusedbytime delaysofthepulsesbetweenelementsofthegroup. Ifamediumincludesechogeneticmatters,thesound scattersonthesurfaceofthemandapartofthescatteringwavecomesbacktothetransducer. Theecho isreceivedbythetransducerthatcanconvertitintoelectricsignals. Themaincomponentofthesignals depends on the frequency characteristic of the transducer and that of the input signal, meaning that, as describedthesignaliscalledRFsignal,thesignalcontainshighfrequencyburstthatcouldnotbedirectly used for a gray scaled image. The image radiologists, physicians, surgeons can diagnose is processed from the RF signal. The processing of the echo signal to B-mode images is the following scheme: (1) 47 3.2 UltrasonicImaging filteringtherawsignalthatisalsocalledradiofrequency(RF)dataineachscanlinetosuppresslowand high frequency components, (2) demodulating the raw signal into enveloped data, and (3) compressing thedatalogarithmically. (1) Filtering Thereceivedsignalcanbefilteredbyhigh-passfiltertocutlowfrequencynoisethatcauseswrong distribution of brightness along the depth direction. It can be also filtered by low-pass filter to remove high frequency noise that arises from digital sampling. The cut-off frequencies of both filters should be lower and higher than the frequency characteristics of the transducer that can be measuredbypulseechotechnique. Thereceivesignalmayincludetheinfluenceofinputburstsin theinitialpartofthetimedomain,whichshouldbecutoff. Therefore,thedurationthatisaffected bytheinputburstscannotbeusedforultrasoundimagesandtheimagescannotshowtheimmediate vicinityofthetransducer. Thefilteringisappliedintimedomainnotinfrequencydomaininmost cases in ultrasound because the Fourier transform is not suitable for realtime imaging that is one of the most advantageous aspects of ultrasound in imaging modalities so that convolution-based filterssuchasfiniteimpulseresponseandinfiniteimpulseresponsefiltersareusuallyimplemented. (2) Demodulation Sofar,theuselesspartofthereceivesignalhasalreadybeenremovedorsuppressedbuttheprocess isnotenoughforustounderstandtheinformationcontainedinthereceivesignalyet. Enveloping the signal can provide us with important information that is really similar to the principle of the radio. The demodulation is the same mechanism as one incorporated with radios. Two kinds of demodulation technique are usually used in ultrasound demodulation: Hilbert transform and IQ demodulation. Theresultsaremathematicallyequaleachother. 48 3.2 UltrasonicImaging Hilberttransform Theanalyticsignaliscalculatedbyfollowingequation: x + (t)=x(t)+ j d x(t); (3.1) where d x(t) is Hilbert transform of x(t) meaning the scan line data. In frequency domain, Hilberttransformmeans: H(f)= 8 > > > > > > < > > > > > > : j (f >0) 0 (f =0) +j (f <0) (3.2) Then,thefrequencyregionoftheanalyticsignalispresented: X + (f)= 8 > > > > > > < > > > > > > : 2X(f) (f >0) X(f) (f =0) 0 (f <0) (3.3) Ifx(t)=m(t)cos(2pft),then d x(t)=m(t)sin(2pft). So,thedemodulatedsignalm(t)canbe foundon: jx + (t)j=jm(t)j: (3.4) 49 3.2 UltrasonicImaging IQdemodulation ThecarrierwaveV C andthesignalV S aredefinedasfollowing: V C = Csin(2pf C t) (3.5) V S = Scos(2pf S t) (3.6) Themodulatedwavecanbedescribedas: V m =(V S +C)sin(2pf C t) =Csin(2pf C t)+ S 2 (2p(f C f S )t)+ S 2 (2p(f C + f S )t) (3.7) And, the baseband In-phase(I) and Quadrature(Q) components of the echo are calculated as followings, I =(V S +C)sin(2pf C t)sin(2pf i t) = V S +C 2 (2p(f C f S )t) V S +C 2 (2p(f C + f S )t) (3.8) Q=(V S +C)sin(2pf C t)cos(2pf i t) = V S +C 2 (2p(f C f S )t)+ V S +C 2 (2p(f C + f S )t) (3.9) where f i is the center frequency of the transmit wave. After filtering I and Q into I ′ and Q ′ respectivelybylow-passfilter,thesecondtermsofEq. 3.8and3.9shouldberemoved. Then, √ I ′ +Q ′ = V S +C 2 : (3.10) thesignalV S isobtained. Inthisstudy,thedemodulationisappliedbyusingHilberttransformbecauseitissimpletoincor- 50 3.2 UltrasonicImaging porate the function of Hilbert transform on MATLAB R ⃝ that is used to generate B-mode images by computing from RF signal. If the demodulation is built in imaging system hardware, IQ de- modulation might be better because the components in electronics needed for IQ demodulation couldbesimplerthanthatforHilberttransform. (3) Logcompression Aftergeneratingtheenvelopesofthereceivesignal,theyarecompressedlogarithmicallybasedon themaximumbrightnesspointinamappedimage,whichcompletesaB-modeimage. Theechoesreturnedfromorgansandothersinthepatient’sbodyareamplifiedbyapreamplifierwith a gain of a few decibels, high input impedance and low noise after transforming the echo signals into electric signals [1]. Next, the time-gain-compensation (TGC) amplifier whose gain is larger than 40dB gains the signals. It compensates the attenuation of ultrasound in medium and tissue. The amplitude of the echoes are also influenced by diffraction of ultrasound that may be made up by TGC. A various kinds of TGC can be applied such as linear gain and exponential gain that amplify the signal in time or depth. Figure 3.11 shows a block diagram of B-mode imaging with a linearly scanned single element transducer. ThisisatypicalsystemofUBMthatcanscantwo-dimensionally. Theultrasoundbeamsent out of a single element transducer is typically focused by lens embedded on the tip of the transducer or by concave tip that is for self-focusing. Then, the echo comes from only a specific region where the focusedbeampenetrates. Thereby,puttingtheenvelopedechosignalssidebysidetoscanningdirection cangenerateB-modeimage,indicatingfigure3.12. Onthecontrary,theultrasoundbeamgeneratedbyanarraytransducerisvariableandprogrammable ofthenumberofelementstransmittingtheultrasound,time-delaythatdefinesfocaldepth,andtheangle of the beam though the focal depth on the aspect of elevation that is perpendicular to azimuth of the probeisfixedbyacousticlenswithrespectto1Darraytransducer,whichiscalledtransmitfocusing. In addition, the array can receive echoes on every channel if each switch is on. The receive signal in time 51 3.2 UltrasonicImaging Figure3.11: BlockdiagramofasingleelementB-modescanner. [1] canbedeterminedwhereallpointsonthesignalcomefrombythedefinitionofscanline,whichiscalled receive focusing. The concept of receive focusing is shown in figure 3.13. This is an example of the receive focusing by using a linear array with 192 scan lines, each of which is produced by 64 channels ofthearray. Mathematical expression of the transmit and receive focusing would like to be shown below. The timedelay∆t n isequaltothedifferenceofthedistancebetweenthenthelementandcenterelementfrom adatapointdividedbysoundspeed. Itcanbesimplifiedto ∆t n = x n sinϕ x c + x 2 n 2cr (3.11) where x n is the distance between nth element and center element, (r,ϕ x ) is the point at which should be imaged,andcissoundspeed. Thomenius[37]summarizedthebeam-formingfunction: e(t)= N å i=1 A ri N å j=1 A tj V ( t∆t ri ∆t tj + 2r(t) c ) (3.12) 52 3.2 UltrasonicImaging Figure 3.12: Schematic of B-mode scanning by using a single element ultrasonic transducer, the image isoriginalfromaclassofBME535orBME536atUniversityofSouthernCalifornia. wheree(t)isthesummedechowaveform,V(t)isthetransmittedwaveform,N isthenumberofarrayel- ements,r(t)isthefocaldistanceataparticulartime,A ri andA tj aretheweightingfunctionsforreception at channel i and transmission channel j respectively, and ∆t ri and ∆t tj are the time delays of reception channel i and transmission channel j respectively. The latest ultrasound imaging system processes all signalsdigitallycalleddigitalimagingsystemarchitectureshowninfigure3.14. The scan line is defined by a single element ultrasonic transducer or array transducer. It takes t f1 53 3.2 UltrasonicImaging Figure3.13: Theconceptofreceivefocusbylineararray secondstomakeonescanline: t f1 =2D=c; (3.13) where D is the depth of penetration determined by the pulse repetition frequency of the pulser and c is the sound speed in tissue. Of course, one image can be made by using the average of N A data lines and thenumberofscanlinesofN s attheexpenseofframerate. Thetimeneededtoformoneframeofimage t f issimplycalculated: t f =2DN A N s =c (3.14) aswellasF =1=t f meanstheframeratethatexcludesthedurationofscanningtransducerineachframe. The duration is quite short in electronic scan. In this study, the center frequency of the needle array is more than 60 MHz that is incompatible to clinically approved systems. For the evaluation of the needle array,utilizingsyntheticapertureimagingtechniqueisreasonable. 54 3.2 UltrasonicImaging Figure 3.14: Block diagram of an ultrasonic imaging system architecture that utilizes a digital beam former. 3.2.2 SyntheticApertureImaging For reducing electric channels due to cost, space and complexity, synthetic aperture imaging can be appropriate. One of the example on the application of synthetic aperture imaging is an intravascular imaging for which the probe should be small and disposable; it is impossible to mount many integrated circuits on the probe [1]. The synthetic aperture imaging does not need to use any huge electronics but simplelinesconnectedtoprocessorandpulser. Althoughonescanlinecanbeacquiredatoncebyusing astandardultrasoundsystemarchitecture,itcanbedonebyaddingsignalsfromthenumberofchannels on the array that are needed in the synthetic aperture imaging technique. One element for transmission 55 3.2 UltrasonicImaging and one element for reception are selected, and voltage pulse from a pulser is inputted to the element for transmission that forms ultrasound. Then the echo coming back to the transducer is received by the selectedelementforreception. Thesignalconvertedbytheelementissenttomemoryafteritisdigitized. Next, 2elementseachfortransmissionandreceptionarenewlyselected. Thisprocedureisrepeatedfor thenumberofthecombinationof2elementsfortransmissionandreception. Afteralldatamemorized,an imageshouldbereconstructedintolikeaB-modeimage. Notonlyforreducingtheelectroniccomplexity butalsoforobtainingimageswithhighcontrastandresolution,thesyntheticapertureimagingtechnique is useful because the advantage of this approach is that a full dynamic focusing can be applied to the transmission and the reception [38]. Figure 3.15 shows the procedure of synthetic aperture imaging that exemplifies it in case all elements receive echo simultaneously [38]. Synthetic aperture imaging by transmissionfrommultipleelementswasalsosuggestedtoobtainhighcontrastimages[39]. Figure3.15: Highresolvedimagecanbereconstructedbysummationofeachlowresolvedimagethatis generatedbyoneelementtransmission. [38] 56 3.2 UltrasonicImaging Figure 3.16: Geometric relation between the transmit and receive element combination and the focal point. [38] The synthetic aperture image reconstruction technique is demonstrated by Trots et al. [38] This method is to calculate the delay of each transmit/receive pair to obtain the signal at each pixel in the image field. For explaining the calculation of this method, the diagram is shown in figure 3.16. To calculatethedelayt m;n betweentransmitandreceiveelementsinthearray,theexpressionis, t m;n = √ x 2 m +r 2 2x m rsinq √ x 2 n +r 2 2x n rsinq c ; (3.15) where x m and x n are the positions of mth element and nth element respectively, (r,q) is the point in the image field shown in figure 3.16. Summed echo signal that is synthetic A-scan signal from each pair of theelementsareprocessedbyfollowingequation: e(t)= N å m=1 N å n=1 y m;n (tt m;n ); (3.16) where y m;n is the received echo signal of each pair. The fact that has to be added is that the frame rate ofthesyntheticapertureimagingismuchlowerthantheultrasoundwithastandardimagingsystemdue 57 3.2 UltrasonicImaging to the repetitive procedure, which seems to be significantly disadvantageous. However, if the standard ultrasound system could be used, because the synthetic aperture technique can provide a image with higher contrast and resolution than the standard ultrasound as mentioned, there may be a lot of rooms wherethesyntheticapertureimagingcanbeapplied. 3.2.3 BackscatteringAnalysis Mostoftheconventionalultrasoundformedicaldiagnosesdemonstratequalitativeimages. Forexample, B-mode image is one of the most prevalent diagnostic modalities but B-mode shows echogeneticity of thebodytissueandradiologistsonlycanspeculatethehealthofthetissueonthebasisoftheirexpertise. Incontrast,itisknownthattissuecharacterizationcanbedonebyextractinginformationtheultrasound from the tissue contains. Ultrasonic tissue characterization is done through untangling hidden patterns in echo data in which the information of tissue function, structure, and pathology must be included [2]. ElastographyandDopplerflowimagingareexamplesofultrasonictissuecharacterization. Backscatter- ingofultrasoundcanbeusedtocharacterizetissues. Aparametercalledbackscatteringcoefficientisoftenusedanditisdefinedasscatteringcross-section dividedbythevolumeofthescattererandsteradians. Thescatteringcross-sectionisdefinedasthepower scattered by the object per unit incident intensity. In case the scatter is a sphere and the radius of it is muchsmallerthenthewavelength,inRayleighscatteringregion,thescatteringcross-sectionisexpressed as[40]: s s = 4pk 4 a 6 9 ( G e G G 2 + 1 3 3r e 3r 2r e +r 2 ) (3.17) wherekisthewavenumber,aistheradiusofthesphere,G e andGaretheadiabaticcompressibilitiesof theparticleandthesurroundingmedium,r e andr arethecorrespondingmassdensities. Thescattering cross-section has a unit of area, then the backscattering coefficient has a unit of per length-steradian. 58 3.2 UltrasonicImaging Therefore,itcanbesaidthatthebackscatteringcoefficientisdependentonfrequencyofultrasound,the size of scatterer, compresibilities, that is to say, stiffness or modulus, and density. Yuan and Shung [41] showedEq. 3.17isanadequateagreementwithexperimentalresultonscatteringbyredbloodcells. IntegratedBackscattering Since the body tissue is fairly inhomogeneous, the backscatter may vary greatly at any single frequencyandoverabandoffrequencies. Hence,theintegratedbackscatteringisfrequentlyused, which is defined as the frequency average of the backscatter over the band width of the signal. O’Donnelletal. [42]andThomasetal. [43]devisedthemethodfortheevaluationoftheintegrated backscatteringcoefficientbyusingthepowerspectrumfromastandardplanereflectorinthefocal zone in order to eliminate the influence of the characteristics except the tissues of interest. The definitionofthe’relative’integratedbackscatteringcoefficientisgivenby: IB= 1 2∆f ∫ f c +∆f f c ∆f jV(f)j 2 jR(f)j 2 df = ∫ +¥ ¥ jv(t)j 2 dt ∫ +¥ ¥ jr(t)j 2 dt (3.18) wherev(t)isareturnechoA-linefromthetargettissuevolumeandr(t)isareferencesignalfrom the standard plane reflector, jV(f)j 2 and jR(f)j 2 are their power spectra respectively, and ∆f is halfthe useablebandwidth. The relationship of the equation between time and frequencydomain is based on Parseval’s theorem. The application with the integrated backscattering technique is represented as 1) cardiac movement monitoring, 2) intravascular ultrasound, and 3) the particle separation. Miller et al. [44] showed that myocardial ultrasonic backscatter is related to the con- tractionalstateofthetissue. Characterizationofthecyclicvariationofintegratedbackscatteringis clearlycorrespondenttothatofcardiacpulsationmeasuredbyelectroorultrasoniccardiologybe- causeofthevariationofmyocardialcomplianceanditsanisotropy. Thetechnologycanbeapplied forthediagnosisformyocardialinfraction,forexample. TheintegratedbackscatteringtechniquehasalreadybeencommercializedbyVolcanoCorporation 59 3.2 UltrasonicImaging (SanDiego,CA)thatproducesultrasoundsystemsforIVUSasshowninfigure1.6. Itisessential to distinguish what kinds of plaque are located on the wall of blood vessel then IVUS with tissue characterizationhasbeendeveloped. Thereasonisbecauseithasbeenhypothesizedthatvulnera- bleplaquesconsistingofalipidcorewithafibrouscaparemostlikelytorupture,whichcanresult in the clot and heart attack. The Volcano Corporation provides the ultrasound that shows several kindsoftissueonthebloodvessel: fibrous,fatty,calcifiedandnecrotictissue[45]. Kawasakietal. [46] clinically investigated the diagnostic accuracy of integrated backscattering IVUS and optical coherencetomographyfordiagnosisofhumancoronaryplaques. Recently, novel applications of high frequency ultrasound have been suggested. One of them is a particle or droplet sorting technology by using ultrasound force. J. Lee et al. [47] and C. Leeetal. [48]developedthemethodofsensingandsortingdropletsbyusingultrasoundforboth. Ultrasonictransducersensestheechotodistinguishthediameterofthedropletbasedonintegrated backscattering coefficients that are sensitive on the size of the droplet. Another application of the integrated backscattering is to detect cell structure. Mercado et al. [49] researched on the estimation of cell concentration by using high frequency ultrasound. These technologies have the potentialtodevelopthetoolformedicaldiagnosis,pharmacologyandbiologicalresearch. SpectralBackscatteringFeatures Theintegrationof backscattering coefficientsin frequencycan reduce strong variancebutit real- izes at the expense of the lost of a lot of information the radio frequency echo signal potentially contains. Feleppa et al. [50], Lizzi et al. [51], and Lizzi et al. [52] proposed the method to char- acterize tissues by using spectral features of backscattering coefficient. They concluded on the spectralbackscatteringfeaturethattheslopeisindicativeofthesizeofthescattererandtheinter- cept is affected by the acoustic concentration, meaning that it is relevant to the scatterer volume concentrationandtherelativeacousticimpedance. Boththeslopeandtheinterceptofthespectral 60 3.2 UltrasonicImaging Figure3.17: Spectralanalysisprocedureofbackscatteringcoefficient. (a)SectionofB-modeimage,(b) logofbackscatteringcoefficientmagnitudevsfrequency. [52] backscattering coefficient are given by fitting the curve of the spectral backscattering features by usingaleast-squaresmethod. Beforethecalculationoftheparameters,thespectralbackscattering coefficientsshouldbecompressedlogarithmicallyasfollowing: BS(f)[dB]=10log 10 jV(f)j 2 jR(f)j 2 (3.19) where a linear fit is applied to BS(f) in order to extract the slope in dB/MHz and 0Hz y-intercept in dB. Figure 3.17 demonstrates the spectral analysis procedure, where L indicating length is de- fined as a window size before computing the spectra. The maps of the slope and y-intercept can beobtainedbythecalculationforeachwindowbyscanningtwo-dimensionally. Thelengthofthe window determines the bandwidth in wave number and the resolution of the maps. Attenuation 61 3.2 UltrasonicImaging shouldbecorrectedbecauseitinfluentstheresultofthespectralfeaturebutitseemstobedifficult tomeasuretheattenuationofactualtissuesduetotheirinhomogeneousmorphology. Feleppaetal. [50]attempted thedistinguishing kinds of tissue in prostate such as benign, hyperplasia, and can- cer. Kolios et al. [54] applied the spectral backscattering technique to analysis of leukemia cells. The apoptosis was observed as the correlation between the slope of the backscattering spectrum andthemean cellradius. Thus, thespectral features of backscattering coefficientscanbe broadly applicableforvariousmedicaldiagnoses. Inthisstudy,theintegratedbackscatteringcoefficientsandthespectralbackscatteringfeatureswereused for characterization of breast biopsy tissue that contains some tissue types such as microcalcification, adenocarcinoma, adipose tissues and fibrous tissues whose diameters of scatterers, acoustic impedance andattenuationaredifferent;thenitappearstobepossibletodistinguishthem. 3.2.4 StatisticalAnalysis Inotherengineeringfieldssuchasmechanical,electrical,chemicalandsoforth,theexperimentalresults should be obviously quantitative and decisive for the conclusion. On the contrary, in biology, medicine, pharmacology and biomedical engineering field, the experimental and clinical results are not always straightforwardly conclusive because the experimental condition is affected by the environment in the tests,thereisnosamebodybesidesgenesthataremadeascloneartificially,andlife-formiscomplexand inhomogeneous. When the experimental results are evaluated, some statistical schemes are introduced toshowwhethertheresultfromtheexperimentconductedforeachgroupisthesameasthatfromother groups or not. The statistical analysis is applied for this study in order to confirm the usefulness of the backscatteringanalysisfordifferentiatingseveraltypesoftissuesinbreastbiopsyspecimens. The performed statistical analysis in this study is described next. For the first step, it should be determinedifthesamplesofthedataderivefromnormallydistributedpopulation: Gaussiandistribution. 62 3.2 UltrasonicImaging Figure3.18: Flowchartofstatisticalanalysis. Atthisstep,Shapiro-WilktestorKolmogorov-Smirnovtest(K-Stest)isapplied. Ifthedataissubjectto sufficiently Gaussian distribution, the procedure progresses to next step. MATLAB R ⃝ providesK-S test soitisadoptedforthefirststep. Next,theF-testisappliedtojudgewhethertwovariableshavethesame variance for the comparison between them with normal distributions. If the result is true, the two are in adequate agreement, then T-test for the judgement if they have the same variance or not. Otherwise, if the result is false, they are not good agreement, Mann-Whitney Rank Sum Test is applied to show if they are significantly different or not. A threshold value of p 0:001 is used for every step of the statisticalprocedureforthedeterminationiftheresultofthetestistrueorfalse. The pistheprobability ofobservingateststatisticasextremeas,ormoreextremethan,theobservedvalueundernullhypothesis, then small p cast doubt on the validity of the null hypothesis. The flowchart of the statistical analysis is 63 3.2 UltrasonicImaging showninfigure3.18. Thisflowwasalsousedin[48],thoughsomekindsoftestsaredifferent. Thetests usedinthisstudyaredescribedbelow,whichisbasicallyaccordingtowebsiteofMathworks R ⃝(Natick MA)[55]. Kolmogorov-Smirnovtest Kolmogorov-Smirnov test as well as called K-S test is a nonparametric test of equality of con- tinuous, one-dimensional probability distribution that can be used to compare a sample with a reference probability distribution (in case: one-sample K-S test), or to compare two samples (in case: two-sample K-S test). In this study, as Gaussian profile is the reference, the one-sample K-S test is applied to show if the data has normal distribution or not. The empirical distribution functionF n isdefinedas: F n (x)= 1 n n å i=1 8 > > < > > : 1 (y i 0) 0 (y i >0) (3.20) TheKolmogorov-SmirnovstatisticforagivencumulativedistributionfunctionF(x)is D n =sup x jF n (x)F(x)j (3.21) The null hypothesis is that the data derive from normal distribution. If p is smaller thana that is of0.001inthisstudy,thenullhypothesiscanberejected,thatis,thedistributionofthedataisnot Gaussianprofile. Inthiscase,thenumberofdatacouldnotbeenough. F-test F-test does not indicate specific test but any statistical tests in which the test statistic has an F- distributionunderthenullhypothesis. F-testismostoftenusedforcomparisonofstatisticalmod- els. The most common example of the use of F-test is that the hypothesis that the means of a 64 3.2 UltrasonicImaging givensetnormaldistributionswhosestandarddeviationsarethesameareequal. Inthisstudy,the functioninMATLAB R ⃝ ’vartest2’isused: F = s 2 1 s 2 2 (3.22) where s 1 and s 2 are the sample standard deviations. The null hypothesis should be rejected if the statisticissufficientlyfarfrom1. Thenullhypothesisinthistestisthatbothstandarddeviationsare same; thenif pislessthana of0.001,thenextstepisMann-WhiteneyRankSumtest,otherwise theproceduremovesontoT-test. T-test T-testaswellascalledStudent’st-testisaparametrictestofthenullhypothesisunderwhenthetest statistic would follow a normal distribution used to determine if two sets of data are significantly different from each other. T-test is generally separated into one-sample test and two sample test. One sample test is used for testing if the average of a set of data is equal to specific value. Two sample test that is used in this study is done for testing if the average values of two groups are equal or not. Under equal sample size and equal variance between two samples, the t statistic to judgeiftheaveragevaluesaredifferentisdefinedas: t = X 1 X 2 s X 1 X 2 √ 1 n (3.23) where s X 1 X 2 = √ s 2 X 1 +s 2 X 2 (3.24) 65 3.2 UltrasonicImaging Ifthesamplesizesarenotequal,thetstatisticisfollowing: t = X 1 X 2 s X 1 X 2 √ 1 n 1 + 1 n 2 (3.25) where s X 1 X 2 = √ (n 1 1)s 2 X 1 +(n 2 1)s 2 X 2 n 1 +n 2 2 (3.26) Iftheaveragesoftwosetsareequal,thestatistict shouldbesubjecttot-distributionofthedegree offreedom: n 1 +n 2 2,whichcanbeanullhypothesis. Thenullhypothesisisrejectedif p<a: 0.001inthisstudy;thenitcanbesaidthattheaveragevaluesofbothsetsaresignificantlydifferent. T-test could be applied if the variances of the sets that should be compared are different but it is notusedinsteadofMann-WhitneyRankSumtestinthisstudy. Mann-WhitneyRankSumtest Mann-Whitney Rank Sum test as well as called Mann-Whitney U test is a nonparametric test of thenullhypothesisthattwosamplescomefromthesamepopulation. ThestatisticU iscalculated as: U =min { n 1 n 2 + n 1 (n 1 +1) 2 R 1 ;n 1 n 2 + n 2 (n 2 +1) 2 R 2 } (3.27) wheren i isthesamplesizeforithgroup,andR i isthesumoftheranksinsamplei. Ifn 1 andn 2 are smallerthan20,checkingMann-Whitneytestchartcanleadtheresult. Forexample,whenn 1 =7 and n 2 =17, U ′ =28 under two-tailed testing, p =0:05. If U is smallter than U ′ , then the null 66 3.2 UltrasonicImaging hypothesisisrejected. Whenn i islargerthan20,theGaussianapproximationisapplied: z = Um u s U (3.28) m U = n 1 n 2 2 (3.29) s U = √ n 1 n 2 (n 1 +n 2 +1) 12 (3.30) wherem U meanstheaverageofU ands meansthestandarddeviation. Theresultisevaluatedby usingnormaldistributionchart. p<a rejectsthenullhypothesis. 67 Chapter4 QuantitativeBackscatteringAnalysisfor CharacterizationofBreastBiopsyTissue 68 4.1 Introduction 4.1 Introduction Ultrasound-guidedbreastbiopsyisprevalentforstandard biopsymethodtoanalyzeabnormaltissuesin breast histopathologically. Percutaneous ultrasound is kind of helpful to locate the target of the biopsy butnumerouspatientswhounderwentthebiopsyreceivedfalsenegativediagnosisthatallowscancerto progress seriously due to the inaccurate position of the biopsy [25]. Cummins [36] proposed a novel biopsy needle on which ultrasonic linear array transducer is mounted, which has the potential of the innovation on the ultrasound-guided breast biopsy. Since radiologists can locate the abnormality from bothoutsideandinside,thenlocalizationoflesionscanbeimproved. Thefrequencybandoftheneedlearrayshouldbe higherthanconventionalultrasoundinclinics be- cause the array needs to obtain the information of tissues in the vicinity of the array, and low frequency region can be covered by conventional ultrasound. The center frequency of the needle array that Cum- minspresentedisaround70MHz[36]. Thehighfrequencyultrasoundhasthecapabilitytocharacterize tissues [2], that is, the needle array can obtain the information of not only the place of the abnormality indetailbutalsothetypesofit. Althoughitiscertainthatpathologyisthemostaccuratetechnologyfor breastcancerdiagnosis,ifradiologistscanlocatelesionsduringbiopsy,nextstepofmedicalprocessfor patients such as operation or medication can be swift, and the number of misdiagnoses can be reduced, whichcanbehelpfulforpreventingcancersfromspreading. In this study, it is shown that high frequency ultrasound has the potential as a tool for tissue charac- terization,specificallyforbreastbiopsytissues. Beforethefabricationofactualneedlearrays,astandard UBMscanningmethodologyisappliedtoprovethefeasibilityofthenovelneedlearrayfortheproposed ultrasound-guided breast biopsy. Concretely, the echo signals from breast biopsy specimens acquired by single element high frequency ultrasonic transducer are analyzed with backscattering analysis, and comparedtohistologicalimages. 69 4.2 Method 4.2 Method Tissue specimens and ultrasound data are commonly used as Cummins et al. [56] experimented. They attemptedtocharacterizethetissuesbyusingNakagamifilteringtechniquewhichenvelopedechosignals were used. In this study, what is different from theirs is to use raw radio frequency (RF) echo signals thatcontainmorequantitativeinformation. Althoughtheycouldberedundantinformationreferringtheir study, since they are fundamental, the tissue specimen and data acquisition are described in this thesis. Theoutlineofthisstudyisshowninfigure4.1. 4.2.1 TissueSpecimenAcquisitionandProcessing Breast core biopsy tissue specimens were acquired at Norris Cancer Hospital and Los Angeles County- UniversityofSouthernCalifornia(LAC+USC)MedicalCenteraspartofthisInstitutionalReviewBoard approved study. These specimens were immediately placed in containers with Dulbecco’s phosphate bufferedsalinesolutionthentransportedtotheResourceCenterforMedicalUltrasonicTransducerTech- nology at USC. Next, each of these specimens was placed in a petri dish and encased in a 3 % agar gel (Fisher Scientific, Waltham, MA) with around 3 mm thick of agar below and surrounding the perimeter of the tissue specimen, leaving the top surface since the ultrasonic transducer was positioned directly above the tissue and any reflection or attenuation from the agar were avoided. The fixation by using the agarwassignificantinordertoaccordtheplaneofultrasoundandhistopathologicalsectioningthatwere conductedafterultrasounddataacquisition. After ultrasound image scan, the ultrasonic transducer on the 3-axis positioning stage was replaced with a razor blade in order to make a plane that is parallel with scanning plane by ultrasound; then ultrasoundandhistologicalimageswereabletobecompared. Theagarblockwith3flatplanethatare2 outersideand1bottomwasthenplacedinacontainerwith10%neutralbufferedformalin(pH6.8-7.2 @ 25 o C) and underwent histopathological sectioning which includied paraffin embedding sectioning at 70 4.2 Method 5mmintervals,andplatingthetissueonglassslidesforhematoxylinandeosin(H&E)staining. ALeica SCN400 digital pathology slide scanner (Leica Microsystems, Inc., Buffalo Grove, IL) then scanned each slide. All data were uploaded to the Leica digital image hub and could be assessed and compared totheultrasoundimages. Figure4.1: Outlineofclinicalstudytoprovetheefficacyofhighfrequencyultrasoundimaginginiden- tifyingcancerinbreastbiopsyspecimens. 71 4.2 Method 4.2.2 TissueSpecimenImagingExperimentalSetup Figure 4.2 illustrates the experimental setup for ultrasonic imaging of breast biopsy tissues. A single element ultrasonic transducer is attached to a 3-axis motorized positioning stage that can determine the position accurately and move things three-dimensionally to arbitral position within the limitation. The pictureofitisshowninfigure4.4whenitisusedformeasurementofpulseechoreflectedbyapolished quarts plate, which the data acquired is used as reference for backscattering analysis. A Panametrics 5900PRpulser/receiverwasusedtosendaunipolarexcitationsignaltothetransducerandthereturnecho signals were gained of 26 dB by it. The signal was captured by a digitizer with a sampling frequency of 1 GHz (Dynamic Signals, Lockport, IL). The images were captured at 100 mm intervals across the width of biopsy specimens, producing a plural of parallel ultrasound images being able to be compared withhistologicalimages,whichisshowninfigure4.3thatisanoverheadview. Thescanlinepitchalong lateraldirectionisdependentonthesizeofspecimen. Figure4.2: Theillustrationofexperimentalapparatusfortheacquisitionofultrasoundechofrombreast biopsytissuespecimensusingsingleelementhighfrequencyultrasonictransducer. 72 4.2 Method Figure 4.3: Biopsy specimens were secured in an agar gel block to maintain their orientation during ultrasoundandhistologicalimaging. Thecharacteristicsofthesingleelementultrasonictransducerusedforthebackscatteringanalysisis shownintable4.1. Theshapeandthestructureofthetransduceristhesameasapictureshowninfigure 3.2andaschematicshownintherightoffigure3.3respectively. The piezoelectric material of the transducer was PMN-PT and was backed by E-Solder 3022 con- ductiveepoxy(VonRollUSA,NewHaven,CT).Thefirstmatchinglayerwasmadeof2-3mmdiameter silver particles (Sigma Aldrich, St. Louis, MO) mixed with Insulcast 501 epoxy (ITW Polymers Coat- ings North America, Montgomeryville, PA). The second matching layer was vapor deposited parylene. The ground connection was made via a Cr/Au electrode plated across the front surface of the 2-3 mm silver epoxy matching layer to the brass transducer housing. An SMA connector threaded into the back ofthebrasshousingcompletingthegroundconnection. Thesignalconnectionthroughthecentersignal port of the SMA connector shorted to the conductive silver epoxy backing of the transducer. This SMA connector then connected to the pulser/receiver via a coaxial cable. The pulse echo data in time and in frequencyareshowninfigure4.5. Thebandwidthof-6dBrelativetothemaximumvalueis27MHzas 73 4.2 Method Figure4.4: ThephotooftheexperimentalarrangementofUBMonthemeasurementofpulseecho. shown in table 4.1 but the transducer is still sensitive in low frequency region. The sensitivity is higher than noise level. Figure 4.6 presents B-mode images to compare with and without low frequency com- ponents: thelowfrequencycomponentinthetopimagewasreducedbyfiltering. Obviously,thedeeper area in the bottom image of the figure is clearer than that in the top one. Additionally, lower frequency componentdeterminestheaccuracyofthebackscatteringcoefficientsextractedfromRFdatawithinthe effectivebandwidth. Therefore, theeffectivebandwidthwasdefinedfrom41MHzto88MHzbasedon thesignallevelof-8dBrelativetothemaximumvalue. 74 4.2 Method Table4.1: Thecharacteristicsofthesingleelementultrasonictransducerusedtoacquireultrasoundecho fromthespecimensofbreastbiopsy. Property Value Material PMN-PT Focus Press-focused CenterFrequency 74MHz BandWidth 27MHz Diameter 2.25mm Focaldepth 3.4mm F-number 1.5 Figure4.5: Thedataofthepulseechoofthesingleelementultrasonictransducerusedforthebackscat- tering analysis: the left shows time response of the pulse echo and the right shows frequency spectrum indBbasedonthepeakvalue. 75 4.2 Method Figure 4.6: The comparison of B-mode images: the top shows low frequency components are cut off andthebottomiswithoutapplicationofthefilter. Thebottomoneshowsrelativelyapparentspecklesin deeperareathatindicatesfattytissue,onthecontrary,itshowsmuchopaqueinthetopone. 76 4.2 Method 4.2.3 QuantitativeBackscatteringAnalysis Although B-mode image is the most popular imaging in ultrasound, it shows only qualitative informa- tion. Ultrasound has more quantitative information that can contribute to analyses and diagnoses for many kinds of diseases such as breast cancers. One of the methods to characterize the tissue specimen isbackscatteringanalysisasdescribedinsection 3.2.3. Integratedbackscattering and spectral backscat- teringfeaturewhicharesubjecttoEq. 3.18andEq. 3.19respectivelycanbeusefulfordeterminationof typesofbreastbiopsytissues. Inthisstudy,theintegratedbackscatteringcoefficientandthespectralbackscatteringcoefficientwere computed by using the RF echo signal detected by the high frequency single element transducer shown in table 4.1. The principle was demonstrated by Lizzi et al. [52]. The pulse echo spectrum of the transducer that has already been shown in figure 4.5 was captured beforehand for the normalization of thebackscatteringcoefficients. Figure4.7: Thetendencyoftheresultofbackscatteringcoefficientinwindowlength. To begin with, RF data scan line was averaged by adjacent lines, that is, 3 lines averaged scan lines were generated for the reduction of the noise. If the RF data were high S/N ratio enough, this would 77 4.2 Method Figure4.8: Theflowchartofthebackscatteringanalysis. be unnecessary. Rightly, the RF data was already filtered to reduce lower and higher frequency noise and cancel the influence of input pulse. The distance of scan lines is small enough related to the lateral resolutionofmorethan30mm. ThepartofRFdatawaspickedwiththelengthofthewindowexemplified in figure 3.17. The length should be longer than 10 times of the wavelength at the center frequency of ultrasound [53]. However, when too long window is used, the window could include multiple kinds of tissues;thedatawouldnotbesuccessfullyevaluatedbecauseofthecomplexity. Theoptimizationofthe length of the window was attempted. Figure 4.7 shows that the data in the range from 200 mm to 450 mmdonotchangebutalmostconvergesothelengthof300mmwasselected. 78 4.2 Method The window scanned the RF signal from the vicinity of the transducer to the end of it. Hanning windowfunctionwasappliedtothepickeddatatomaketheperiodicityofthedatafornextstep: digital Fouriertransform. AnyotherwindowfunctionsuchasHammingwindowandBlackmanwindowarealso available. AfterprocessingtheFouriertransformofthedata,thiswasdividedbythereferencespectrum: the pulse echo of the transducer, for normalization. Eventually, spectral backscattering coefficient was obtained. ApplyingEq. 3.18,theintegratedbackscatteringcoefficientwasextracted,andapplyingleast- square approximation after the logarithmical compression: Eq. 3.19, the slope and the y-intercept of the spectral backscattering coefficient were calculated. Then, these parameters were 2-dimensionally mapped but these maps had a lot of variation. To remove the spiky variations in the slope and the y- interceptmaps,spatiallow-passfilterthatneverchangemainvalueoftheparameterswereappliedalong theaxialdirection. Theflowoftheanalysisisillustratedinfigure4.8. Aftercompletingthecalculation, thecomparisonbetweentheultrasoundimagesandthehistologicalimageswasperformedtoidentifythe locationatwhicheverytypeoftissueinthebreastbiopsyspecimensexists. 4.2.4 QuantitativeandStatisticalAnalysis In this study, statistical analyses for the explanation of the validity with respect to the result of the backscattering analysis were performed, which can determine whether there were any statistically sig- nificant differences between tissue types in breast biopsy specimens. The method and the procedure of thestatisticalanalysiswasalreadyshowninsection3.2.4. Thenumberofslicesbeingabletocompareto histologicalimagesis51intotalthatconsistsof4cases,meaning4specimensthatwerebiopsiedfrom4 differentpatientswhohaveinvasivebreastcancer. Becausethetissueconditionsweredifferentbetween 4casesandbecauseitisimpossibletoobtainsameorsimilartissuesselectively,thenumberofslicesin eachspecimenandthetypesoftissuethespecimenscontainaredifferent. 5pointsoneachtissuesuchas fibrous tissue and adipose tissue were picked from each slice of ultrasound but the number of points of microcalcifications depends on how many parts of microcalcifications each slice has. Spatial averaged 79 4.2 Method valueinregionofinterest(ROI)wereextractedtoevaluatethebackscatteringcoefficientsinsteadofthe value at one point. The size of the ROI is 250mm250mm, but that for each microcalcification that is typically smaller than 100mm is 50mm50mm. The example of setting ROI is shown in figure 4.9. Average value and standard deviation of each kind of the backscattering coefficients were calculated by usingtheseROIs. Figure 4.9: The example of setting the region of interest for statistical analysis of backscattering coeffi- cients. 80 4.3 ResultandDiscussion 4.3 ResultandDiscussion Theresultoftheanalysisincomparisonbetweenquantitativeultrasoundandpathologyofbreastbiopsy specimens indicates the backscattering coefficient may be able to differentiate tissue types. One of the cases shows that adenocarcinoma that is cancerous was distinguished from adipose tissues: fat cells by integrated backscattering coefficient and spectral backscattering feature. Microcalcifications in some cases were significantly differentiated from other tissues around them. However, the inter-specimen comparison was not successful, that is, the backscattering coefficient from the same type of tissue did notshowgoodagreementwitheachother. Thepossiblereasonsofthediscrepancywillbeshowninthis section after the quantitative results of this analysis. One of things is the number of specimens is too small to obtain any meaningful conclusion; therefore, the result of this analysis can only suggest that the backscattering analysis has the potential or one of the candidates to distinguish abnormality from normal tissues. In the following discussion, each set of the backscattering parameter maps for cases 1 - 4willbeshownwiththecorrespondingB-modeimagesandH&Estainedhistologicalimages. Thenthe statisticalanalysisresultsforcase1-4willbedescribedhowabnormaltissuesuchasmicrocalcification andadenocarcinomainbreastwereabletobedistinguishedfromothersineachspecimen. The4breastbiopsyspecimenswereanalyzedbyusingbackscatteringtechniquesandhistopathology. The basic information of the specimens is shown in table 4.2. It shows the sex and ages of patients, the number of slices of ultrasound images, tissue types of which each specimen contains as well as the diagnoses of cancer. As shown in table 4.2, the number of slices of each specimen is not the same because of different size between specimens. This is inevitable in clinical study. The biopsy method is different, for example, case 1 was biopsied by 14-Gauge core biopsy needle but case 4 was obtained by 11-Gauge vacuum assisted stereotactic biopsy, that can be dependent on the location of the legion. In thisstudy,tissuetypesthatthespecimenscomposearenotconsistentthrougheachspecimenbutinclude mostoftypicaltissuetypesinbreast. 81 4.3 ResultandDiscussion Table4.2: Thebasicinformationofthespecimens. Case Sex Age No. ofSlices TissueType Diagnosis Case1 Female 50 9 Adenocarcinoma,Adipose invasiveductalcarcinoma Case2 Female 58 14 Adipose,Fibrous, metastaticadenocarcinoma Microcalcification Case3 Male 68 5 Adenocarcinoma poorlydifferentiated adenocarcinoma Case4 Female 59 23 Fibrous,Microcalcification invasiveductalcarcinoma 4.3.1 2DMapsofBackscatteringCoefficient In total 51 ultrasound-histology image sets were generated from 4 breast biopsy specimens as shown in table 4.2. A set of co-registered 2D maps of ultrasound backscattering coefficients was drawn to presenttheextentoftheusefulnessforthecharacterizationoftissuetypesandmicrostructuresinbreast. The image sets in addition to B-mode image and histological image will be shown as the integrated backscattering coefficient in dB, the slope of the spectral backscattering coefficient in dB/MHz, and the y-intercept of it in dB. The values in these 3 parameters are depicted in a color scale with blue and red representingthelowandhighvaluesrespectively. Case1: The tissue in case 1 was biopsied by using 14-Gauge core biopsy needle. The result of pathology indi- catesthetissueclearlycomposesof2types: adiposetissueandadenocarcinoma. Theboundaryofthe2 typesarerelativelyobviousaccordingtobothultrasoundandhistologicalimage. 9ultrasound-histology image sets were created from the specimen and figure 4.10 presents one of them. Two kinds of circles are depicted on every image on the set shown in figure 4.10, which indicate adipose area is marked by red oval and adenocarcinoma area is done by green oval respectively. B-mode image of the set barely tells the difference between the two types of tissues but backscattering parameters clearly separate both of them. 5 regions of interest in each type are picked from each slide in order to calculate the statistics 82 4.3 ResultandDiscussion of backscattering parameters and to analyze how statistically different the backscattering coefficients of thebothtissuesare. Whatshouldbeaddedisthatthelocationofthespecimenwasclosertofocaldepth thanthatoftheothercases;whereROIswerepickedtobeasclosetofocusaspossible. Table4.3: Meanandstandarddeviationofbackscatteringparametersincase1. TissueType IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Adipose 66:23:4 0:250:07 52:67:0 Adenocarcinoma 58:23:2 0:0680:070 55:53:4 Theaveragedandstandarddeviationofbackscatteringcoefficientsoneachtissueareshownintable 4.3. According to the table 4.3, the difference of integrated backscattering values are about 8 dB and the slope of spectral backscattering coefficient of adipose tissue is significantly different from that of adenocarcinoma, which is also obvious by observing the 2D maps in figure 4.10. 0Hz-intercept of both arenotsodifferentfromeachother. Basedonthemapofthebackscatteringcoefficientandthestatistics of it, it is suggested that the adipose tissue and the adenocarcinoma can be distinguished by using the backscatteringanalysis. Case2: The tissue in case 2 was obtained from a patient diagnosed with metastatic adenocarcinoma consistent with a breast primary. The specimen separated into 2 pieces at the time of biopsy; then the ultrasound including backscattering images are segmented into Part 1 (left side) and Part 2 (right side) as shown in figure 4.11 which is one of 14 ultrasound-histology image sets of the specimen. From the histological image in figure 4.11 (e) and (j), several microcalcifications are identified as well as adipose and fibrous tissues but no adenocarcinoma was observed from the specimen. The red arrows points the position of microcalcifications,andtheredcirclesandthegreencirclesindicatefibroustissueareaandadiposearea, respectively. 83 4.3 ResultandDiscussion (e)Histologicalimage Figure 4.10: Case 1 image set includes high frequency ultrasound. The red dashed circles indicate adiposetissueareasandgreencirclesindicateadenocarcinoma. 84 4.3 ResultandDiscussion Table4.4: Meanandstandarddeviationofbackscatteringparametersincase2. TissueType IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Adipose 43:23:2 0:380:08 22:04:6 Fibrous 43:52:9 0:240:09 30:66:2 Microcalcification 34:85:0 0:310:22 17:512 The B-mode and backscattering images clearly indicate the location of microcalcifications because they are hyperechoic. In this case, they which are smaller than 100 mm were detected. The relative acoustic impedance of microcalcification is larger than that of other tissues or media such as water. Nonetheless, the microcalcifications are not detected by using conventional ultrasound because of the acoustic attenuation and the resolution of ultrasound; therefore, the images suggest that high frequency ultrasoundisapplicableforbreastcancerdiagnosis. Whenlookingatthemapsoffigure4.11moreclosely,integratedbackscatteringandy-interceptval- uesonthepositionwheremicrocalcificationspresumablyexistindicatehighvaluesthoughtheechogenic- ity of the scatterer is a function not only of acoustic impedance which depends on density and sound speed or modulus but also of the size of the scatterer besides of ultrasound wave length. In short, the echogenicityofmicrocalcificationdependsoneachsizeofit. Theevidenceofthisisshownintable4.4. The standard deviation of backscattering parameter of microcalcification is larger than that of other tis- sues. Anyway,itcanbesaidthatmostofsmallhyperechoicspotsinbreasttissuearemicrocalcifications. With respect to other types of tissue such as adipose and fibrous tissue, the mean and the deviation values of them are close. The possible reasons are followings: the specimen has mixed parts of adipose and fibrous tissues. In fact, the fibrous tissue area where was defined contains small but many adipose cells. And, the shape of tissue in ultrasound cannot be maintained during the pathology; therefore, it mightbedifficulttoselectthelocationonspecifictissueaccurately. Anotherreasoncouldbethatfibrous andadiposetissuecouldnotbeseparatedbyusingultrasoundbecausetheirechogenicitymightbeclose. This investigation could be future work because a large number of clinical investigation will have to be 85 4.3 ResultandDiscussion Part1 Part2 (e)Histologicalimage (j)Histologicalimage Figure4.11: Case2imagesets: leftrowispart1andrightrowispart2,includehighfrequencyultrasound. The red arrows indicate microcalcifications and the red and green circles indicate fibrous tissue and adiposetissuerespectively. 86 4.3 ResultandDiscussion done to achieve the conclusion, fibrous tissue and adipose tissue are not related to cancerous tissue in breastbiopsythough. Case3: The specimen in case 3 was obtained from a male patient diagnosed with poorly differentiated adeno- carcinoma consistent with a breast primary. A pathologist who diagnosed this specimen remarked that thespecimenonlyconsistedofadenocarcinoma. Oneoftheultrasound-histologysetsisshowninfigure 4.12. 5regionsofinterestwerepickedfromeachsliceandthenumberofslicesofultrasoundinthiscase is5;then25regionsofadenocarcinomawerestatisticallyanalyzed. Themeanandstandarddeviationof backscattering coefficients are shown in table 4.5. The lateral length of the image is longer than that of othercases,andthepitchofscanlinesof25mmislargerthanthatofothersof10mm. Table4.5: Meanandstandarddeviationofbackscatteringparametersincase3. TissueType IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Adenocarcinoma 49:02:4 0:240:06 36:65:0 The B-mode and backscattering images indicates the values are not sharply variable, which is the evidencethatthetissueinthisspecimenisalmosthomogeneous,indicatingthetissuecontainsonlyone typeoftissue: adenocarcinoma. Case4: Thespecimenincase4wasbiopsiedwithan11-Gaugevacuumassistedbiopsyneedleunderobserving mammogram called stereotactic breast biopsy. The patient was diagnosed with invasive ductal carci- noma, which was concluded by using another biopsy sample that was obtained during the same biopsy procedure, but the specimen in case 4 does not contain any tumors. The ultrasound-histology image set isdisplayedinfigure4.13onwhichsomearrowsareaddedtopointatseverallocationsofmicrocalcifi- 87 4.3 ResultandDiscussion (e)Histologicalimage Figure 4.12: Case 3 image set includes high frequency ultrasound. The histological image shows the biopsyspecimencomposesofadenocarcinoma. 88 4.3 ResultandDiscussion cations. TheB-modeimageshowsthatareasbehindmicrocalcificationsarehypoechoic,whichindicates microcalcifications tend to reflect acoustic waves backward and to prevent them from propagating for- ward. Table4.6: Meanandstandarddeviationofbackscatteringparametersincase4. TissueType IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Fibrous 50:53:7 0:0670:071 47:94:5 Microcalcification 43:13:8 0:160:13 35:19:6 Thenumberofultrasound-histologysetsis23. Thetissuecontainsfibroustissuebesidesmicrocalci- fications. So115ROIsoffibroustissueandthenumberofmicrocalcificationsthatareabletobepicked are used to analyze backscatter coefficients statistically. The mean and standard deviation of them are shown in table 4.6. Integrated backscattering and y-intercept of spectral backscattering coefficients can differentiatebetweenmicrocalcificationsandfibroustissuethatsurroundseachmicrocalcificationinthis specimenduetohighacousticimpedanceofmicrocalcificationsasobservedinCase2. 89 4.3 ResultandDiscussion (e)Histologicalimage Figure4.13: Case4imagesetincludeshighfrequencyultrasound. Theredarrowsindicatethepositions ofmicrocalcifications. 90 4.3 ResultandDiscussion 4.3.2 StatisticalAnalysisResults In addition to mapping ultrasound backscattering coefficients as demonstrated in previous section, sta- tistical analyses that are generally important to support the conclusion of the investigation related to biology, medicine and other fields was performed on those data. In this study, the statistical analysis in each case was conducted by the scheme according to the section 3.2.4 described previously. Box plots that are typically used to show the analysis of statistics were applied for each case in order to support the result of the analysis and to assist one to grasp them. All boxplots shown below are defined: red midline in each box is median value, bottom edge and top edge of the box indicate the 25th and 75th percentiles,respectively,thewhiskersextends2.7s thatisequivalentthat99.7%datacontainsbetween thewhiskers,andtheoutliersareplottedusingthe’+’symbol. Case1: Box plots in figure 4.14 show the distribution of integrated backscattering coefficient, the slope and the y-intercept of spectral backscattering features for the case 1. All parameters of backscattering shown in figure 4.14 indicate that the difference of them between adipose tissue and adenocarcinoma was found tobestatisticallysignificant. Basedonthestatisticalprocedureshowinginfigure3.18,theyarenormaldistributionbytheresultof K-Stest(p>0:2). Andallofthemfailedtopassequalvariancetest. Mann-Whitneytestconductedfor all of them. As results, although p values of the intercept of the spectral backscattering feature is large, whichdoesnotshowstatisticaldifferencebetweentwogroups, pvaluesoftheintegratedbackscattering coefficient and the slope of the spectral backscattering features are less than 0.001; the difference be- tween the two groups is greater than would be expected by chance, which means there is a statistically significantdifference. 91 4.3 ResultandDiscussion Figure 4.14: Box plots for integrated backscattering coefficient (top), slope (middle), and y-intercept (bottom)ofadiposeandadenocarcinomaincase1. 92 4.3 ResultandDiscussion Case2: Figure4.15showstheboxplotsforbackscatteringcoefficientsforthecase2. Someparametersareclose between types of tissues whereas others are clearly different. Look at the top of figure 4.15 that shows theboxplotsoftheintegratedbackscatteringcoefficients. Thisparameterofadiposeandfibroustissues are close that cannot be differentiated each other but that of microcalcification is clearly different from theothers. Aswellastable4.4,thedeviationonmicrocalcificationislargerthanothers. There are 3 combinations that should be compared in statistics. Before the statistical analyses, K-S test resulted in the confirmation that all distributions are normal. The first one is between adipose and fibrous tissue. Besides the integrated backscattering coefficient that did not show significant difference betweenadiposeandfibroustissue,theslopeandtheinterceptofspectralbackscatteringfeaturesdiffer- entiatedbothtissues(p<0:001)accordingtotheresultsofMann-Whitneytestintheslopeandt-testin theintercept. Thenthesearestatisticallydifferentratherthanwouldbeexpectedbychance. Next,thebackscatteringcoefficientsfrommicrocalcificationandfibroustissuewerecomparedbyus- ingthesameprocedure. Mann-Whitneytestsforalltheparametersindicatethattheintegratedbackscat- tering coefficient and the intercept of the spectral backscattering feature of both are from the different origins(p<0:001)thoughitcannotbesaidthattheslopeofthemcomefromdifferentgroup(p=0:01). Therefore,thetwogroupsarestatisticallymoredifferentthanwouldbeexpectedbychance. Finally,thebackscatteringcoefficientsfrommicrocalcificationandadiposetissueswerecomparedin statistics. All p values of Mann-Whitney tests for all the backscattering parameters are less than 0.001; thedifferencebetweentwogroupsaresignificantlydifferentinstatistics. Intheend,allofthethreegroupsarestatisticallydifferentbasedontheperformanceofthestatistical analyses. 93 4.3 ResultandDiscussion Figure 4.15: Box plots for integrated backscattering coefficient (top), slope (middle), and y-intercept (bottom)ofadipose,fibroustissueandmicrocalcificationsincase2. 94 4.3 ResultandDiscussion Figure 4.16: Box plots for integrated backscattering coefficient (left), slope (center), and y-intercept (right)ofadenocarcinomaincase3. Case3: Box plots showing the distribution of backscattering coefficients for the case 3 are depicted in figure 4.16. Sincethisspecimenhadonlyadenocarcinoma,nocomparativeanalysiswasperformed. Case4: The distribution of backscattering coefficient on fibrous tissue and microcalcification for the case 4 is plotted in figure 4.17. The echogenicity of microcalcifications is larger than other tissue around them sothedifferenceisobviousbutthestatisticalanalysiswasperformed. Allbackscatteringparametersfor both fibrous tissue and microcalcification are normal based on the results of K-S tests (p>0:1). Inte- grated backscattering coefficients failed to pass the equal variance test but the slope and the y-intercept ofthespectralbackscatteringfeaturebetweenbothtypesoftissuepassed. Forthem,Mann-Whitneytest conductedwhereast-testconductedtheintegratedbackscatteringcoefficientfortheintegratedbackscat- tering coefficient. As results, all of p values are less than 0.001; the difference between the two groups is greater than would be expected by chance, which means the two groups are significantly different in statistics. 95 4.3 ResultandDiscussion Figure 4.17: Box plots for integrated backscattering coefficient (top), slope (middle), and y-intercept (bottom)offibroustissueandadenocarcinomaincase4. 96 4.3 ResultandDiscussion 4.3.3 AdditionalDiscussion The backscattering analysis should be for the extraction of the characteristics of each tissue, in other words,theobjectiveofbackscatteringshouldbethecomparisonofinter-specimens. However,according to the result of all cases presented above, the values of backscattering coefficients of the same types of tissue do not show any good agreement. Of course, actual tissues are really complex, which is one of the reason but there still remains what should be considered: attenuation and pressure distribution that shouldinfluencethebackscattering. Attenuation It is known that the acoustic attenuation in tissue is mostly proportional to frequency [1]. This is the reason why high frequency ultrasound is unavailable in conventional medical ultrasound. The region in frequencyinthisstudyisstronglyaffectedbytheattenuation;thenitshouldbeconsideredforthetissue characterization. Fortunately,thespecimenincase1comprisesofrelativelyhomogeneousadiposetissue andadenocarcinoma,whichcanprovidetheattenuationcoefficientsofthem.Thespectralradiofrequency datawerepickedat z=2:75mm,3.0mm,3.25mm,and3.5mminordertocalculatetheattenuationin frequency,averagingwithhundredsscanlinesineachtissuearea. Theratiooftheaveragedspectraldata between different depths indicates the attenuation. Even averaged, a carve of the attenuation coefficient in frequency is not simple but sometimes looks ugly. Then, the linear approximation being subject to following: a 0 = 1 dz ln { V(z) V(z+dz) } = 1 dz ln { 1 s 1 f +s 0 } ; (4.1) wasapplied, wherea 0 istheattenuation coefficientin nepers per centimeter,dz is the distance between two measured points, V(z) is a radio frequency echo signal at z in frequency domain, f is frequency in MHz, and s 0 and s 1 are coefficients for the linear approximation. Figure 4.18 shows the attenuation 97 4.3 ResultandDiscussion coefficientsofadiposetissueandadenocarcinomaincase1. Itindicatesthatthedifferentiationbetween adiposetissueandadenocarcinomacanbecausedbythedifferenceoftheattenuationcoefficients. Then the compensation of the backscattering coefficients by using the attenuation coefficients was attempted. Theresultisshownintable4.7,indicatingbothareclosebesidesslightdifferenceonintegratedbackscat- teringcoefficientandy-interceptofthespectrum. Figure4.18: Attenuationcoefficientincase1. Table4.7: Meanandstandarddeviationofbackscatteringparametersincase1,whichiscompensatedby usingtheattenuationcoefficientextractedfromtheradiofrequencyechosignal. TissueType IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Adipose 57:12:3 0:0770:082 64:05:4 Adenocarcinoma 54:33:5 0:0700:078 60:63:7 Although the attenuation mainly contains scattering and absorption in tissue, the scattering can be ignored [58]; then this means the difference of absorption between the two kinds of tissues. Anyway, 98 4.3 ResultandDiscussion the backscattering analysis is still useful to distinguish adenocarcinoma from adipose tissue by using broadbandultrasonictransducerandbroadbandinputpulseduetothedifferenceofattenuation. The attenuation coefficient of other tissues and other samples was attempted. The result is shown in figure 4.19. The coefficient on adenocarcinoma in case 3 gives similar values to that in case 1 and that on fibrous tissue in case 2 is much larger than others. Normal tissues such as fat and glandular tissue attenuate sound wave rather than carcinoma [59]. However the value shown in figure 4.19 may notbesoaccuratebecauseofthepressuredistributionfromthetransducerandcomplexityofthetissue. Thecompensationofthebackscatteringcoefficientfortheattenuationwasattemptedforinter-specimen comparison. The result is shown in table 4.8 and 4.9, indicating the discrepancy of the values between specimensremains. Figure4.19: Attenuationcoefficientsofseveralkindsoftissues 99 4.3 ResultandDiscussion Table 4.8: Mean and standard deviation of backscattering parameters of adipose, which is compensated byusingtheattenuationcoefficient. Case IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Case1 57:12:3 0:0770:082 64:05:4 Case2 40:43:0 0:260:06 26:04:3 Table4.9: Meanandstandarddeviationofbackscatteringparametersofadenocarcinoma,whichiscom- pensatedbyusingtheattenuationcoefficient. Case IntegratedBackscattering(dB) Slope(dB/MHz) y-Intercept(dB) Case1 54:33:5 0:0700:078 60:63:7 Case3 47:42:5 0:160:05 39:04:7 DistributionofPressure The other reason why inter-specimens evaluation cannot succeed in this backscattering analysis is the highly focused transducer was used so the profile of pressure sharply vary in space. Figure 4.20 shows the pressure profile along axial direction and spatial distribution that is simulated by using Field II. Then,thetissuepositioncouldbesensitivetotheresultofthebackscatteringanalysis. Forexample,the specimen in case 1 was situated on 2 mm <z< 3.5 mm, the specimen in case 2 was on 3.5 mm <z< 4.5 mm, and that in case 3 was laid on 3.0 mm <z< 4.0 mm. These should have been tweaked to the samelocationaslongaspossible. Thecompensationbyusingthesimulationresultwasattemptedbutit wasnotsuccessfulbecausetheexperimentalpressureprofilemaynotbeideal. 100 4.3 ResultandDiscussion Figure4.20: Beamprofileofthesingleelementtransducerusedinthisstudy. Leftimageis2Dprofileof thepressuregeneratedbythetransducer,rightimageshowstheaxialprofile. 101 4.4 Conclusion 4.4 Conclusion In this study, the backscattering analysis of breast biopsy tissues obtained from breast cancer patients was performed by using single element ultrasonic transducer with higher frequency than is clinically used. The analysis demonstrated that microcalcifications that are invisible in low frequency ultrasound wereclearlyvisualizedanddifferentiatedfromothernormaltissuesbyusinghighfrequencyultrasound. Thisobservationsuggeststhatthecapabilitytodetectmicrocalcificationsofthisdeviceissignificantfor theimprovementofthebreastcancerdiagnosis. Backscatteringcoefficientsfrommicrocalcificationsare also so conspicuous that high frequency ultrasound can clearly differentiate them. In addition, adeno- carcinomawasshowntobedistinguishablefromadiposeinthisstudy. Theseresultsshowtheproposed devicemayholdpromisetoimprovetheaccuracyofdiagnosticbreastpathologyandtoreducethenum- berofpatientswhowouldreceivefalsenegativediagnoses. However, thisstudyshouldbestrengthened byadditionalresearchandseveralimprovementsoftheexperimentalcondition,whicharegiven: 1. Acquiremorenumberofsamplesandmoredata 2. Chooseappropriatetransducersforthebackscatteringanalysis 3. Bemorecarefulinselectingthelocationoftissueonwhichthemeasurementofthebackscattering coefficientswillbecarriedout. Thesamplesize: fourspecimensistoosmallnumbertoyieldanymeaningfulconclusion;therefore, additional sampling should be done as future work. At least 20 specimens are necessary in order to confirm the experimental result statistically. The pressure profile of the transducer used in this study is very short because it is self-focused. Data acquired by a non-focused transducer or a linear array transducermaybebettercompensatedtocalculatethebackscatteringcoefficientmoreaccuratelytoallow inter-specimen comparison. Transducers with broader bandwidth are better for backscattering analysis, especially for the spectral features because it would be more accurate to extrapolate y-intercept and to approximatetheslope. 102 Chapter5 BuildingTestingSystemforHigh FrequencyUltrasonicTransducerArray 103 5.1 Introduction 5.1 Introduction As mentioned in section 4.4, high frequency ultrasound has the potential to differentiate or characterize tissue types in breast. At UTRC, a novel breast biopsy needle has been developed [36]. The needle is integrated with a miniature high frequency ultrasound array that can obtain ultrasound images from the needleinthevicinityofit,whichcanimprovetheaccuracyofpositioningthebiopsywithpercutaneous ultrasound guidance. Ultrasound-guided breast biopsy is a gold standard diagnostic method but the in- accuracyoftissuesamplingleadsfrequentfalsenegativediagnoses[26]. Highfrequencyultrasoundcan resolvemicrostructuresuchasmicrocalcificationthatispossibletobeasignofprecancerouschangesor early breast cancers but ultrasound attenuates through breast and the attenuation is approximately pro- portional to frequency. This is the reason why high frequency percutaneous ultrasound is not available tovisualizethemicrostructureinbreast. Therefore,thedevelopmentofhighfrequencyultrasonictrans- ducer integrated in a breast biopsy needle is extremely worthwhile that can contribute to the reduction ofthemortalitycausedbybreastcancerbecausethedevicecanhelpradiologistslocalizelesionssuchas DCIS. Highfrequencyultrasonictransducerarrayisdifficulttobeevaluatedbyperformingimagingexper- imentbecausetheconventionalcommercializedimagingsystemiscompatibletolessthan20MHz. Our laboratorydevelopedthesystemforhigherfrequencybutthatworksatmost37.5MHzbasedonthesam- pling frequency of 150 MHz [57]. The frequency of the array integrated in the needle should be higher than that: from 60 to 80 MHz. At UTRC, the imaging system for high frequency will be developed but thereexistsnomethodatpresentfortheevaluationoftheneedlearray. Thensyntheticapertureimaging technique that can evaluate the array without any complex and expensive electronics is introduced. In this study, the objective is not only the evaluation of the transducer array on the needle but building up the evaluation system for any high-frequency arrays. The system had to be able to acquire the signals fromallchannelsofthearrayandtoreconstructanimageimmediatelyaftertheacquisition. 104 5.2 Method 5.2 Method 5.2.1 SyntheticApertureImagingSystem The evaluation system is concisely illustrated in figure 5.1. It composes of ultrasonic transducer array that has multiple elements: 64 elements in case with the needle array, transmission lines connecting each element of the array with PC-Borad that contains RF connecter and transformer for each line, and a multiplexer with both transmit port and receive port. PC controls a pulser for transmission and data acquisition system (DAQ) for reception of echo. A LabVIEW (National Instruments, Austin, TX) program running on a PC operates the multiplexer that is a modular NI PXIe 2593 multiplexer system (National Instruments, Austin, TX), shown in figure 5.2. This is a configurable multiplexer platform that uses computer controlled mechanical switches to channel high voltage signals within the 500 MHz bandwidth. Each PXIe 2593 module is configured into a 28 multiplexer to provide separate transmit and receive channels for 8 individual array elements. 8 of these modules are combined into 2 separate rows of 8 transmit and 8 receive channels are connected, which creates a 264 multiplexer unit that enablesthesystemtotransmitandreceiveindependentlyonall64elementsoftheultrasoundarray. This isacoreofsyntheticapertureimaging. Thesystemcantransmitandreceiveonindependentchannelsandechosignalisacquiredfromeach of the 6464 =4096 transmit/receive pairs. Each element is excited by applying a high voltage and short pulse, and the echo is received with each element, which results in 4096 transmit/receive signals that are for imaging by using synthetic aperture imaging technique. This technique is demonstrated by Trotsetal. [39]andalsoexplainedinsection3.2.2. 5.2.2 SoftwareforSyntheticApertureImaging A custom built LabVIEW software program that controls the system that allows users to test any trans- ducer arrays and imaging algorithms were also developed. The program controls RF data acquisition, 105 5.2 Method Figure5.1: Imagingsystemblockdiagram. multiplexer channel switching, data processing and image reconstruction. It has a graphical user inter- face(GUI),asshowninfigure5.3. ThroughtheGUI,theuserscanselecttheelementsofthearraythey desiretotestandthetypeofimagescanning. TheGUIprovidesuserswiththeflexibilityoftesting. The GUIwasdevelopedinordertouseforversatilekindsofhighfrequencyultrasoundimaginginthefuture. UserscanusetheGUIasfollowings: (1) First, a user chooses the mode: "Recording Data" and/or "Display Image" both of whose buttons areoval-shaped,locatedontop-leftoftheGUI.If"RecordingData"isturnedon,theprogramisto acquirethedata. Iftheuserturnsononly"DisplayImage",itgeneratesanimageusingpre-stored data. If both are on, the both programs are done sequentially. If no button is turned on, the list of 106 5.2 Method Figure5.2: Themultiplexerprovidesindependentswitchingbetweentransmitandreceivechannels. theelementsisonlycreatedandsaved. (2) Second,theuserinputs"Startelem.","Endelem.",and"pitch",whicharefundamentalinformation of the array that will be evaluated. Scan setting is selected from "Fully Sampled", "Same Tx & Rx",and"Manual",whichareexplainedontable5.1. (3) Beforestartingtheprogram,theuserhastoinputthedatafoldername,otherwisethedatamaybe overwritten. Afterthestartispressed,theLabVIEWstartstocontrolthemultiplexerandDAQfor eachselectedpair: transmit/receive. (4) Duringdataacquisition,thenumberiscountedup. Oncethedataacquisitioncomplete,thiscount upisstopped. If"DisplayImage"isnotpressed, thedataare savedinthe filethatwasdesignated in "Data Folder", then the program ends. If "Display Image" was turned on, it is noticed that 107 5.2 Method Figure5.3: GraphicuserinterfaceprogrammedinLabVIEWforsyntheticapertureimaging. 108 5.2 Method Table5.1: ScantypesinSyntheticApertureImaging. ScanSetting Description FullySampled A pulse/echo waveform is captured from each transmit-receive ele- ment pair in the array. Each waveform is used for the synthetic aper- tureimagereconstructionprocessproducingaB-modeimage. SameTx&Rx Apulse/echowaveformiscapturedforeacharrayelementwithasin- gle element acting as both the transmit and receive transducer. These echowaveformsaretreatedasscanlinesandarrangedtogethertoform aB-modeimage. Manual(ReadScanList) This option enables the user to define any scan list for the array. The multiplexer can independently select transmit and receive elements, makingimagedataacquisitionflexible. Usersdefineallthesequential transmit-receive element pairs in an table list file. This file is then uploaded to the LabVIEW program and used to control the scanning process. the signal is shown in bottom of the display. It provides the user with the information about the selection of the axial range of an image that should be reconstructed. The user decides and input "Depth_start"and"Depth_end". Iftheuserinputsthe"Depth_end"thatisover"max"theusercan notice on the right of "Depth_end" box, the loop keeps going until the end of the axial position is properlyselected: the"Depth_end"hastobeselectedlessthan"max". (5) Press "Go Imaging" after selecting the axial range. Immediately after pressing, at most a few minutes,theimageisreconstructedandpresented. Ofcourse,thewaitingtimeisdependentonthe number of channels and scanning type that was selected. The image is automatically saved with MATLAB R ⃝format. Each sequence of the data acquisition consists of the initialization of the multiplexer, the selection the pair elements for transmission and reception, and the acquisition of the RF data in DAQ with the multiplexer. Thetriggerpulse thatsynchronizesDAQandfilter is generatedby the pulser. AftertheRF 109 5.2 Method dataisacquired,LabVIEWreadsthedatathataredigitizedinDAQfromonboardmemoryoftheDAQ. Andthen,thesequencerepeatedlycontinuesandreadssubsequentrowofthescanlistuntilitapproaches theendofthelist. An original synthetic aperture imaging algorithm was coded by using MATLAB R ⃝ . LabVIEW has MATLAB R ⃝ script box so the MATLAB R ⃝ code could be directly embedded on the sequence of Lab- VIEW. Although this was attempted, the routine of imaging became too long to complete the imaging. Then, some parts of code of synthetic aperture imaging were re-written in LabVIEW code instead of MATLAB R ⃝ because the time-consumption was attributed to MATLAB R ⃝ script box provided by Lab- VIEW.Afterthismodificationofthecode,thetimeoftheimagingbecameshorttoatmostafewminutes. 5.2.3 MiniatureUltrasonicTransducerArray Figure 5.4: Transducer on the glass with transmissionlines. Figure5.5: Zoomedviewofthetransducer. Aminiatureultrasonictransducerarraythatwasusedinthisstudywillbedescribed. Thedetailofit was already published by Cummins [36]; hence, only a summary of the transducer should be described in this section. The transducer is for the integration in a breast biopsy needle for which 11 AWG is assumed. Then, the transducer sample was integrated in a needle-like-shaped glass, which is shown in 110 5.2 Method figure 5.4. On the tip of the glass, the transducer array that has 64 elements is glued and electrically connected to transmission lines on the glass, shown in figure 5.5. The array is unfocused or naturally focused in elevation direction. The layer structure of the transducer is viewed as an exploded image in figure5.6. Table5.2showstheparameterandmaterialoftheacousticstacks. Thepiezoelectricmaterial wasetchedbyusingdeepreactiveionetching(DRIE).Oneelementconsistsoftwenty50mm-lengthsub- elements. The top viewof the etched piezoelectric material pictured with scanning electron microscope isshowninfigure5.7. Otherlengthofsub-elementswereattemptedbutthefabricationofthe50mmwas thebestin allattempts. Thediced chipwith the transducer is bonded on the needle glass byconducting epoxy bumps on each edge of element, which is also electrically connected. The opposite of the tip of theneedlewasadaptedtoPCboardwithtransformersandconnectorswiththemultiplexer. Table5.2: Acousticstackproperties. Layer Material Thickness Density V long V shear AcousticImpedance (kg=m 3 ) (m/s) (m/s) (MRayls) Matching Parylene a 7mm 1100 2350 1662 2.6 Piezoelectric PMN-PT b 24mm 8100 4608 3258 37.0 KerfFiller Epo-Tek301 c 24mm 1150 2650 1270 3.0 Backing E-Solder3022 d 1mm 3200 1850 1308 5.9 Thespecificationofthetransducerisshownintable5.3. AccordingtoCummins[36],thearrayhad noshortedoropenelements,however7elementshadverylowsensitivitywithinsertionlossvaluesover -60 dB. A typical result of the pulse echo is shown in figure 5.8, which indicates the central frequency is 59.6 MHz and -6 dB fractional bandwidth is 34.7 %. The overall variation of center frequency and fractional bandwidth of this array is shown in figure 5.9. The measured crosstalk was -23.7 dB, which is because the distance between adjacent elements in the 2-2 composite is very narrow and because a ONDACorporation,Sunnyvale,CA b HCMaterials,Bolingbrook,IL c EpoxyTech.,Billerica,MA d VonRollUSA,NewHaven,CT 111 5.2 Method Table5.3: Themeasurementspecificationoftheminiaturetransducer. Property Value Numberofelements 64 Sizeofelement 1mm14mm Pitch 12mm Kerfwidth 6mm Piezomaterialthickness 22mm Centerfrequency 59.1MHz Fractionalbandwidth 29.4% Sensitivity 703mV Insertionloss -41.0dB Electricalimpedancemagnitudeat60MHz 408.4Ω Focalpoint 2.2mm electrical cross-coupling between element electrodes is high due to mutual electrical impedance of the elements. Thevariationandcrosstalkshouldbereducedforobtainingbetterqualityoftheimage. Experimentalsetupofthesyntheticapertureimagingwillbedescribed,anditisthesameasthepulse- echoevaluationofthetransducerthatwasdonepriortoperformingtheimaginginordertodeterminethe effective bandwidth, sensitivity, pulse length, and focal distance of the transducer. The both of the tests wereperformedinde-ionizedwater. Figure5.10presentstheexperimenttoimageawirephantom. The pulser used in pulse-echo test was a Panametrics 5900PR pulser/receiver (Panametrics Inc., Waltham, MA).Theamplitudeofthepulseisappliedof100V pp . Toreceiveechosignals,abandpassfilterpassing the frequency range of 10 to 100 MHz and a gain of 26 dB were applied before the GaGe digitizer (DynamicSignalsLLC.,Lockport,IL)thatisalsoandataacquisition(DAQ)withthesamplingrateof1 GHzdigitizedanalogechosignals. Thetargetofthesyntheticapertureimagingwasfirstlyasingle20mmtungstenwire(CaliforniaFine WireCompany,GroverBeach,CA).Thewirewasset1.5mmapartfromthetransducer. Forthetest,an AVTechAVB2-C-USCCMonocyclePulseGeneratorwasusedtogeneratea70MHzsinglecyclepulse withanamplitudeof160V pp . 112 5.2 Method Figure 5.6: An exploded array view of the miniature ul- trasonictransducer. Figure5.7: Scanningelectronmicroscope image of PMN-PT material directly after DRIEprocessing. Figure5.8: Pulse-echomeasuredresultsforatypicalarrayelement: (left)echosignalintimeand(right) thatinfrequencyindB. 113 5.2 Method Figure5.9: Centerfrequencyandbandwidthvaluesforeachelementinarray. Figure5.10: Experimentalconfigurationoftheneedlearrayforimagingawirephantom. 114 5.3 ResultandDiscussion 5.3 ResultandDiscussion Imaging for the evaluation of the miniature transducer array was performed by using the GUI as men- tionedpreviously. TheB-modeimagebasedonthesyntheticaperturetechniquewassuccessfullyrecon- structed as shown in figure 5.11. The image reconstruction did not include any thresholding, averaging, or apodization. From the B-mode data, the axial and lateral resolution were calculated. This measured full-width half-maximum (FWHM) axial and lateral resolution were 32.2 mm and 115.6 mm, respec- tively. The line spread functions for the axial and lateral resolution plots for the center of the wire are showninfigure5.12. Accordingtobothfigures5.11and5.12,artifactsaround-20dBareobservedthat seemstobebecausethesensitivityofseveralelementsisnotsufficientlyhightoeraseit. Thebrightness ofanechoicareaisaround-40dBthatisrelativelyhigh. Thereasonistheratioofsignaltonoiseisvery lowduetothemethodofexcitation: ultrasoundgeneratedfromonlyoneelementforeveryecho. In this trial, one selected element was excited and one selected element received the echo, and all combinations of the element pairs: transmission and reception, were performed and all echo data were synthesized. A single echo did not have high amplitude enough to create a large contrast relative to noise level. To improve the contrast, multiple elements should be excited simultaneously, which can be nextstepbutmakingseveralimprovementsofthemultiplexerandprogramsthatcontrolitarenecessary. Another issue of the synthetic aperture imaging is that it is time-consuming because 4096 switching channelsandacquiringdataareneededtogeneratejustoneimage,whichtakesaround1hour. Asanext step, imaging multiple wire phantoms in order to evaluate the contrast and resolution in depth should be done, and imaging tissue samples by using this synthetic aperture technique and clinical studies by using the needle array should be also executed. Besides clinical studies, these tests appear not to be difficultbutthereisanotherproblemoftheneedlearraythatpiezoelectricmaterialofPMN-PTiseasily to be depolarized during testing. The actual main reason of the depolarization is unclear but one of the possible reasons is that the Curie temperature of PMN-PT is lower than 100 o C and the piezoelectric 115 5.3 ResultandDiscussion STA Image 0 0.2 0.4 0.6 0.8 1 1.2 Lateral Distance (mm) 0.8 1 1.2 1.4 1.6 1.8 2 2.2 Axial Distance (mm) -40 -35 -30 -25 -20 -15 -10 -5 0 Figure 5.11: Reconstructed synthetic aperture image of a single 20 mm wire target. The B-mode image wasdisplayedonalineargrayscaleindBrelativetomaximumintensity. 116 5.3 ResultandDiscussion -0.3 -0.2 -0.1 0 0.1 0.2 0.3 Axial Distance (mm) -40 -35 -30 -25 -20 -15 -10 -5 0 Relative Magnitude (dB) -0.3 -0.2 -0.1 0 0.1 0.2 0.3 Lateral Distance (mm) -40 -35 -30 -25 -20 -15 -10 -5 0 Relative Magnitude (dB) Figure5.12: Axialandlaterallineplotsforthecenterofthewirephantom. 117 5.4 Conclusion materialcanbeheatedeasilybyappliedvoltagepulses. Ontheevidenceofit,thesensitivityofthearray wasgraduallydecreasedafterseveraltesting. Thisisacriticalissueforthedevelopmentoftheneedleas acommercialproduct. The software for synthetic aperture imaging developed here at present is only for this miniature linear array, which means only the linear array can be evaluated by using this software. If the software is developed to be compatible to other testings, some have to be added and arranged. For example, curvedarrayandphasedarrayimagingcouldbeimplementedifthesyntheticapertureimagingforthem are coded, and the matching list that the arrangement of the array is coordinated with the address of multiplexerorcircuitsshouldbereadyforeacharray. TheGUImayhavetobeslightlyadjustedforthese arrays or images and some input parameters may have to be added. To be more efficient, programable circuitsusedforamultiplexermaybebetterthanthemechanicalswitchingmultiplexerthatwasusedin thisstudy,whichalsorequirenewprogramsforthecontrolofthecircuits. 5.4 Conclusion In this study, the evaluation system for high-frequency ultrasonic transducer was successfully built up by using LabVIEW control both signal acquisition system and image processing program with graphic userinterface. Theperformanceofaminiaturehigh-frequencyultrasonictransducerarraythathasbeen developedtobeintegratedinabreastcorebiopsyneedlewasevaluatedusingtheevaluationsystem. The systemcanmeasurepulseechoofeachelementandgenerateB-modeimage. Thesignalacquisitionsys- tem composes of pulser/receiver, amplifier, data acquisition system and multiplexer, which can transmit wavefromanelementofthearrayandreceivetheechoatanotherone. Allthecombinationsofreceived data are synthesized by the image processing part of this evaluation system. The synthetic aperture imaging technique was introduced for the evaluation of the high frequency ultrasonic transducer array embeddedwithinthebiopsyneedle. Theprogramofthesyntheticapertureimagingwasincorporatedin 118 5.4 Conclusion LabVIEW, which resulted in the success of building up the system; the performance of the transducers can be judged shortly after the data acquisition is completed. The remained issues of the problem are followings: 1. longtimetoacquirethedatathatareneededtoreconstructtheimage, 2. signal-noiseratioislowerthanactualimagingsystem,whichcauseslowcontrastimages. ThesearegoingtobesolvedbyintroducingtheASICdevelopedmultiplexerandthetransmissionby multipleelementsthatgeneratelargerpressurewavethansingleelementtransmission. Theimagingsys- temcompatibletohigherfrequencyenoughshouldbedevelopedtoevaluatethetransducersandimaging performanceultimately,whichcancontributetosounddevelopmentofhighfrequencyultrasoundappli- cationsanddevices. 119 Chapter6 Conclusion 120 6.1 Summary 6.1 Summary Theresearchconductedinthisthesisisapartofthedevelopmentofminiaturehighfrequencyultrasonic transducer integrated on a breast biopsy needle for the proposed ultrasound-guided breast biopsy which has researched in UTRC at University of Southern California. Two things were mainly demonstrated: 1. backscattering analysis of breast biopsy tissues by using single element high frequency ultrasonic transducerand2. settinguptheevaluationsystemforhighfrequencyultrasonictransducerarrays. First, this study was to show proof of the concept of how high frequency ultrasound is useful for the improvement of the breast cancer diagnosis. It has been known that high frequency ultrasound can resolvemicrostructures such as microcalcifications that can be suggestive of early breast cancer such as ductal carcinoma in situ (DCIS). Here, the high frequency ultrasound performed the detection of mi- crocalcifications. In addition, backscattering coefficient coming from microcalcifications differentiated from other normal tissue quantitatively. Moreover, adenocarcinoma that is one of the malignant cancer- oustissuescanbedifferentiatedfromadiposebyusingbackscatteringcoefficientduetothedifferenceof the attenuation: adipose attenuates acoustic wave rather than adenocarcinoma. Therefore, the backscat- teringanalysisinhighfrequencyultrasoundcanbeasignificanttooltodetectabnormalityduringbreast biopsy,fortheresultsofthediagnosticbreastpathologycanbemoreaccuratebyimprovingtheaccuracy of positioning the breast biopsy needle and reducing false negative results. This study introduced the conclusion that high frequency ultrasound from the breast biopsy needle can be significantly useful and beneficialforthebreastcancerdiagnosis. Second, the evaluation system was successfully built for the high frequency ultrasonic transducer arraywithhigherfrequencythanisclinicallyusedorcompatiblewiththesysteminourlaboratory. The center frequency of the needle array should be more than 60 MHz so we needed to develop not only transducers but also the evaluation systems. The data acquisition system had been developed but the analysisoftheacquiredsignalcouldnotbedoneimmediatelyaftertheacquisition. UtilizingLabVIEW 121 6.2 FutureWork withsplendidGUI,thealgorithmofsyntheticapertureimaginginLabVIEWwasincorporatedtorealize theevaluationofsuperhighfrequencyultrasonictransducers. TheB-modeimageofsinglewirephantom was performed using this cord. It took only a few minutes to reconstruct a linear scanned gray scale image after the data acquisition. The system can be adjusted to any kind of transducer such as phased array if the program and the channel address map are changed. The system can contribute to prompt developmentofanyhighfrequencyultrasonicdevicesandapplications. In summary, the both parts in this study assisted research and development on the miniature trans- ducer array on breast biopsy for the novel ultrasound-guided breast biopsy. Furthermore, this study showed the potential of high frequency ultrasound in medicine through the clinical study. The study is also applicable for the research on other topics regarding high frequency ultrasound. Thus, this study is meaningfulfortheresearchanddevelopmentofhighfrequencyultrasoundinmedicine. 6.2 FutureWork Despitetheachievementinthisstudymentionedintheprevioussection,importantfutureworksstillre- mainsinregardstoboththebackscatteringanalysisandthebuildingsystemforevaluation. Thenumber ofthespecimensinthisstudyisverysmallsoadditionalspecimensofbreastbiopsytissueshouldbeob- tainedandthedatashouldbeacquiredinordertosupportoftheusefulnessabouttheminiatureultrasonic transducerandtheproposedultrasound-guidedbreastbiopsy. Itisalsohopefulthatthiskindoftheclin- ical study should be done by using the high frequency linear array. Regarding the evaluationsystem for highfrequencyultrasound,thedataacquisitiontakesstilllongtime,andthesignalnoiseratioisnotlarge enough. Introducing a programmable multiplexer and excitation on plural elements of the array can be helpfultoacquiretheechosignalfasterandtoreconstructimageswithhighercontrastthanthoseofthe current system. Doing these works should lead us to the realization of the proposed ultrasound-guided breastbiopsyandothercutting-edgetechnologiesinreferencetohighfrequencyultrasound. 122 Reference Reference [1] K. K. Shung (2015) Diagnostic Ultrasound: Imaging and Blood Flow Measurements, 2nd ed. Abingdon, UK:CRCPress. [2] T.L.Szabo(2014),DiagnosticUltrasoundImaging: Insideout,2ndedition.AcademicPress. [3] H.G.Klein(1981),AreB-scanners’sDaysNumberedinAbdominalDiagnosis?,DiagnosticImaging,Vol. 3,pp.10-11. [4] J. Ophir, I. Cespedes, H. Ponnekanti, Y. Yazdi and X. Li (1991), Elastography: A Quantitative Method for imagingtheElasticityofBiologicalTissues.UltrasonicImaging,Vol.13,pp.111-134. [5] F.L.Lizzi,M.Astor,E.J.Feleppa,M.Shao,A.Kalisz(1997),StatisticalFrameworkforUltrasonicSpectral ParameterImaging.UltrasoundinMedicineandBiology,Vol.23,pp.1371-1382. [6] D.A.Knapik,B.Starkoski,C.J.PavlinandF.S.Foster(2000),A100-200MHzUltrasoundBiomicroscope. IEEETransactionsonUltrasonics,FerroelectricsandFrequencyControl,Vol.47,pp.1540-1547. [7] T. M. Blodgett, C. C. Meltzer and D. W. Townsend (2007), PET/CT: Form and Function. Radiology, Vol. 242,pp.360-385. [8] K.R.Nightingale,M.L.Palmeri,R.W.NightingaleandG.E.Trahey(2001),OntheFeasibilityofRemote Palpation Using Acoustic Radiation Force. Journal of the Acoustical Society of America, Vol. 110, pp. 625-634. 123 Reference [9] U. Alqasemi, H. Li, A. Aguirre and Q. Zhu (2012), FPGA-based Reconfigurable Processor for Ultrafast Interlaced Ultrasound and Photoacoustic Imaging. IEEE Transactions on Ultrasonics, Ferroelectrics and FrequencyControl,Vol.53,pp.1344-1353. [10] H. Tanaka, T. Azuma, S. Machida, K. Hashiba and T. Kobayashi (2008) Beam structure for CMUT with desiredfrequencyspectrum.UltrasonicsSymposium.IUS2008.IEEE [11] P.Helin,P.Czarnecki,A.Verbist,G.Bryce,X.RottenbergandS.Severi(2012)MicroElectroMechanical Systems(MEMS),2012IEEE25thInternationalConference [12] A.Bhuyan,J.Choe,B.Lee,I.O.Wygant,A.Nikoozadeh,O.Oralkan,B.T.Khuri-Yakub(2013)Integrated Circuits for Volumetric Ultrasound Imaging With 2-D CMUT Arrays. IEEE Transactions on Biomedical CircuitsandSystems,Vol.7,No.6,pp.796-804. [13] J. M. Cannata, J. A. Williams, L. Z. Zhang, C. H. Hu, K. K. Shung (2011) A High-frequency Linear UltrasonicArrayUtilizinganInterdigitallyBonded2-2Piezo-composite.IEEETransactionsonUltrasonics, FerroelectricsandFrequencyControl,Vol.58,No.10,pp.2202-2212. [14] R. Chen, N. E. Cabrera-Munoz, K. H. Lam, H. Hsu, F. Zheng, Q. Zhou, K. K. Shung (2014) PMN-PT Single-crystalHigh-frequencyKerflessPhasedArray.IEEETransactionsonUltrasonics,Ferroelectricsand FrequencyControl,Vol.61,No.6,pp.1033-1041. [15] T.Cummins, P.EliahooandK.K.Shung(2015)HighFrequencyUltrasoundArrayforUltrasoundGuided BreastBiopsy.(pendingpublishment) [16] J. Yin, H. C. Yang, X. Li, Q. Zhou, C Hu, j. Zhang, K. K. Shung, Z. Chen (2010) Integrated Intravascular Optical Coherence Tomography (OCT) - ultrasound (US) imaging system. Proc. SPIE 7557, Multimodal BiomedicalImagingV,75570D(February23,2010). [17] C.Lee,J.S.Jeong,J.Y.Hwang,J.Lee,K.K.Shung(2014)Non-contactMulti-particleAnnularPatterning and ManipulationwithUltrasound Microbeam. AppliedPhysicsLetters, Vol.104, No. 24, art. 244107, pp. 1-3 [18] M.K.Shetty(2015)BreastCancerScreeningandDiagnosis.Springer. 124 Reference [19] A. N. Sencha, E. V. Evseeva, M. S. Mogutov and Y. N. Patrunov M. K. Shetty (2013) Breast Ultrasound. Springer. [20] http://www.breastcancer.org/: Breastcancer.org. [21] http://www.cdc.gov/: CentersforDiseaseControlandPrevention [22] http://www.cancer.gov/: NationalCancerInstituteattheNationalInstitutesofHealth [23] http://ganjoho.jp/public/ : Center for Cancer Control and Information Services at National Cancer Center (ofJapan). [24] M.H.Gail,L.A.Brinton,D.P.Byar,D.K.Corle,S.B.Green,C.Schairer,J.J.Mulvihill(1989)Projecting Individualized Probabilities of Developing Breast Cancer for White Females Who Are Being Examined Annually.JNatlCancerInst.Vol.81,No.24,pp.1879-86. [25] E.B.Claus,N.RischandW.D.Thompson(1994)AutosomalDominantInheritanceofEarly-onsetBreast Cancer.ImplicationsforRiskPrediction.Cancer.Vol.73,No.3,pp.643-651. [26] J. H. Youk, E.K. Kim, M. J. Kim, J. Y. Lee, K. K. Oh (2007) Missed Breast Cancers at US-guided Core NeedleBiopsy: HowtoReduceThem.Radiographics,Vol.27,No.1,pp.79-94. [27] AmericanCancerSociety(2015),CancerFactsandFigures2015,AmericanCancerSociety,2015. [28] W. Bruening, K. Schoelles, J. Treadwell, J. Launders, J. Fontanarosa and K. Tipton (2009) Comparative Effectiveness of Core-needle and Open Surgical Biopsy for the Diagnosis of Breast Lesions. Agency for HealthcareResearchandQuality. [29] H.Schatten(2013)CellandMolecularBiologyofBreastCancer.HumanaPress. [30] R.delReBrun(2010)MinimallyInvasiveBreastBiopsies.Springer. [31] H.E.Kaiser,A.NasirandT.J.Yeatman(2014)MolecularPathologyandDiagnosticsofCancer.Springer. [32] W.P.Mason(1964)Physicalacousitcs(Vol.1A).NewYork: AcademicPress. [33] R. Krimholtz, D. Leedom, and G. Matthaei (1970) New Equivalent Circuits for Elementary Piezoelectric Transducers.Electron.Lett.Vol.6,pp.398-399. 125 Reference [34] S.Saitoh,T.Takeuchi,T.Kobayashi,K.Harada,S.Shimanuki,andY.A.Yamashita(1999).3.7MHzphased array probe using 0.91Pb (Zn 1=3 Nb 2=3 )O 3 - 0.09PbTiO 3 Single Crystal. IEEE Transactions on Ultrasonics, FerroelectricsandFrequencyControl,Vol.46,pp.414-421. [35] T. A. Ritter, T. R. Shrout, T. Tutwiler, and K. K. Shung (2002) A 30-MHz composite ultrasound array formedicalimagingapplications.IEEETransactionsonUltrasonics,FerroelectricsandFrequencyControl, Vol.49,No.2,pp.217-230. [36] T.Cummins(2015)HighFrequencyUltrasoundArrayforUltrasound-GuidedBreastBiopsy,DoctoralDis- sertationatUniversityofSouthernCalifornia. [37] K. E. Thomenius (1996) Evolution of ultrasound beamformers. IEEE Ultrasonics Symposium, Vol. 2, pp. 1615-1622. [38] I. Trots, A. Nowicki, M. Lewandowski (2009) Synthetic transmit aperture in ultrasound imaging. Archives ofAcousticsVol.34,No.4,pp.685-695. [39] I.Trots,A.Nowicki,M.Lewandowski,andY.Tasinkevych(2010)Multi-ElementSyntheticTransmitAper- tureinMedicalUltrasoundImaging.ArchivesofAcousticsVol.35,No.4,pp.687-699. [40] P.M.MorseandK.U.Ingard(1968)TheoreticalAcoustics.NewYork: McGraw-Hill. [41] Y. W. Yuan and K. K. Shung (1988) Ultrasonic backscatter from flowing whole blood. I. Dependece on ShearRateandHematocrit.JournaloftheAcousticalSocietyofAmerica,Vol.84,pp.52-58. [42] M. O’Donnell and J. Miller (1981) Quantitative Broadband Ultrasonic Backscatter: An Approach to Non- destructiveEvaluationinAcousticallyInhomogeneousMaterials.JournalofAppliedPhysicsVol.52,No.2, pp.1056-1065. [43] L.J.ThomasIII,B.Barzilai,J.E.Perez,B.E.Sobel,S.Wickline,J.G.Miller(1989)QuantitativeReal-time Imaging of Myocardium Based on Ultrasonic Integrated Backscatter. IEEE Transactions on Ultrasonics, FerroelectricsandFrequencyControl,Vol.36,No.4,pp.466-470. [44] J. G. Miller, J. E. Perez, S. A. Wickline, S. L. Baldwin, B. Barzilai, V. Davila-Roman, R. J. Fedewa, A. E. Finch-Johnston, C. S. Hall, S. M. Handley, F. D. Hockett, M. R. Holland, A. Kovacs, G. M. Lanza, S. 126 Reference H. Lewis, J. N. Marsh, J. Mobley, D. E. Sosnovik, R. L. Trousil, K. D. Wallace, and K. R. Waters (1998) Backscatter Imaging and Myocardial Tissue Characterization. IEEE Ultrasonics Symposium, 1998. Vol.2, pp.1373-1383 [45] http://www.volcanocorp.com: VolcanoCorporation,SanDiego,CA. [46] M. Kawasaki, B. E. Bouma, J. Bressner, S. L. Houser, S. K. Nadkarni, B. D. MacNeill, I-K. Jang, H. Fujiwara,andG.J.Tearney(2006)DiagnosticAccuracyofOpticalCoherenceTomographyandIntegrated BackscatterIntravascularUltrasoundImagesforTissueCharacterizationofHumanCoronaryPlaques.Jour- naloftheAmericanCollegeofCardiology,Vol.48,No.1,pp.81-88. [47] J. Lee, J. H. Chang, J. S. Jeong, S-Y. Teh, and K. K. Shung (2011) Backscattering Measurement from a Single Microdroplet. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control., Vol. 58, No.4,pp.874-879. [48] C. Lee, J. Lee, H. Kim, S. Teh, A. Lee, I. Chung, J. Park, and K. K. Shung (2012) Microfluidic Droplet SortingwithaHighFrequencyUltrasoundBeam.LabChip,Vol.12,pp.2736-2742. [49] K.P.Mercado,M.Helguera,D.C.Hocking,andD.Dalecki(2014)EstimatingCellConcentrationinThree- Dimensional Engineered Tissues Using High Frequency Quantitative Ultrasound. Annals of Biomedical Engineering,Vol.42,No.6,pp.1292-1304. [50] E. J. Feleppa, A. Kalisz, J. B. Sokil-Melgar, F. L. Lizzi, T. Liu, A. L. Rosado, M. C. Shao, W. R. Fair, Y. Wang, M. S. Cookson, V. E. Reuter, and W. D. W. Heston (1996) Typing of Prostate Tissue by Ultrasonic Spectrum Analysis. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, Vol. 43, No. 4,pp.609-619. [51] F. L. Lizzi, E. J. Feleppa, S. K. Alam, and C. X. Deng (2003) Ultrasonic Spectrum Analysis for Tissue Evaluation.PatternRecognitionLetters,Vol.24,pp.637-658. [52] F. L. Lizzi, S. K. Alam, and S. Mikaelian, P. Lee, and E. J. Feleppa (2006) On the Statistics of Ultrasonic SpectralParameters.UltrasoundinMedicineandBiology,Vol.32,No.11,pp.1671-1685. [53] K. A. Topp, J. F. Zachary, and W. D. O’Brien Jr. (2001) Quantifying B-mode Images of In Vivo Rat Mam- maryTumorwithFrequencyDependenceofBackscatter.J.UltrasoundMed.,Vol.20,pp.605-612. 127 Reference [54] M. C. Kolios, G. J. Czarnota, M. Lee, J. W. Hunt, and M. D. Sherar (2002) Ultrasonic Spectral Parameter CharacterizationofApoptosis.UltrasoundinMedicineandBiology.Vol.28,No.5,pp.589-597. [55] Mathworks R ⃝website.http://www.mathworks.com/ [56] T.Cummins,C.Yoon,H.Choi,P.Eliahoo,H.Kim,M.W.Yamashita,L.J.Hovanessian-Larsen,J.E.Lang, S.F.Sener,J.Vallone,S.E.Martin,andK.K.Shung(2015)High-frequencyUltrasoundImagingforBreast CancerBiopsyGuidance.JournalofMedicalImaging.Vol.2,No.4,047001. [57] B. Kang (2015) High-Frequency Ultrasound Array-Based Imaging System for Biomedical Applications. DoctoralDissertationatUniversityofSouthernCalifornia. [58] K. K. Shung and G. A. Thieme (1993) Ultrasonic Scattering by Biological Tissues. Boca Raton, FL: CRC Press. [59] F. T. D’Astous and F. S. Foster (1986) Frequency Dependence of Ultrasound and Backscatter in Breast Tissue.UltrasoundinMedicineandBiology,Vol.12,No.10,pp.795-808. 128

Abstract (if available)

Abstract High frequency ultrasonic transducers integrated in a breast biopsy needle have been researched at the University of Southern California. It can obtain information in the vicinity of the needle and resolve microcalcifications that are indicative of precancerous changes or early breast carcinomas such as ductal carcinoma in situ. X-ray mammography is the best diagnostic modality to detect microcalcifications currently, but patients are exposed to ionizing radiation. More recently, a new ultrasound-guided breast biopsy technique was proposed. This technique utilizes conventional ultrasound guidance coupled with high frequency ultrasound from the biopsy needle. Current ultrasound-guided breast biopsy can often result in false negative diagnosis that causes serious consequences because of the inaccuracy of the position of the biopsy needle. This proposed technology can improve localization of lesions and the accuracy of the diagnosis of breast pathology. ❧ In this study, for the purpose of demonstrating the proof of the concept with respect to the miniature high frequency ultrasonic transducer embedded within a biopsy needle, backscattering analysis for the characterization of breast biopsy tissues was conducted by using a single element high frequency ultrasonic transducer with a center frequency of 74 MHz. The effective bandwidth was from 41 MHz to 88 MHz and was defined on - 8 dB relative to the maximum value of the pulse echo spectrum on the transducer. The slope and the 0 Hz y-intercept of the spectral backscattering features were calculated with linear approximation as well as the integrated backscattering coefficient. Four breast biopsy tissue specimens in each agar block used for embedding were measured utilizing a standard ultrasonic bio-microscopy (UBM) scanning methodology. All of the backscattering coefficients were two-dimensionally mapped and co-registered with each histopathological image for the identification of tissue types. ❧ The results of the analysis indicate that the integrated backscattering coefficient successfully differentiated microcalcifications from other tissues around them. And adenocarcinoma is also differentiated from adipose tissues by observing the integrated backscattering and the slope of the spectral backscattering feature. These results indicate that backscattering analysis is able to quantitatively distinguish tissues into normal and abnormal, which should help radiologists locate abnormal areas during the proposed ultrasound-guided breast biopsy with high frequency ultrasound. ❧ Current imaging systems for conventional ultrasound are available for less than 20 MHz. To evaluate the miniature high frequency ultrasonic transducer, a test system with synthetic aperture technique was built, and used to process the data to produce ultrasound images without any large scale electronics. The GUI controlling the data acquisition and the image processing program are beneficial for those who research high frequency ultrasonic transducer arrays. It takes a few minutes to reconstruct a synthetic B-mode image after the acquisition of the RF echo signals. ❧ In conclusion, the backscattering analysis shows that high frequency ultrasound is significantly useful for the proposed ultrasound-guided breast biopsy, and the testing system based on a GUI should be flexibly available for the evaluation of every high frequency transducer array. Therefore, the development of the miniature high frequency ultrasonic transducer integrated in a breast biopsy needle can be easily advanced to the next stage.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

High frequency ultrasound array for ultrasound-guided breast biopsy

PDF

Synthetic aperture imaging platform based on CMOS high voltage 1 to 64 multiplexer / de-multiplexer for ultrasound guided breast biopsy needle

PDF

Design and development of ultrasonic array transducers for specialized applications

PDF

High-frequency ultrasonic transducers for photoacoustic applications

PDF

Development of high-frequency (~100mhz) PZT thick-film ultrasound transducers and arrays

PDF

Development of high frequency focused transducers for single beam acoustic tweezers

PDF

Microfluidic cell sorting with a high frequency ultrasound beam

PDF

Development of front-end circuits for high frequency ultrasound system

PDF

High frequency ultrasonic phased array system and its applications

PDF

High-frequency ultrasound array-based imaging system for biomedical applications

PDF

Single-cell analysis with high frequency ultrasound

PDF

Miniature phased-array transducer for colorectal tissue characterization during TEM robotic surgery; and, Forward-looking phased-array transducer for intravascular imaging

PDF

Array transducers for high frequency ultrasound imaging

PDF

A high frequency array- based photoacoustic microscopy imaging system

PDF

Multi-modality intravascular imaging by combined use of ultrasonic and opticial techniques

PDF

Quantification of cellular properties using ultra-high frequency single-beam acoustic tweezer

PDF

Intravascular imaging on high-frequency ultrasound combined with optical modalities

PDF

Development of novel 1-3 piezocomposites for low-crosstalk high frequency ultrasound array transducers

PDF

High-frequency ultrasound imaging system with Doppler features for biomedical applications using 30~35 mHz linear arrays and an analog beamformer

PDF

Highly integrated 2D ultrasonic arrays and electronics for modular large apertures

Asset Metadata

Creator Akiyama, Takahiro (author)

Core Title High frequency ultrasonic imaging for the development of breast biopsy needle with miniature ultrasonic transducer array

School Viterbi School of Engineering

Degree Master of Science

Degree Program Biomedical Engineering

Publication Date 04/19/2016

Defense Date 04/18/2016

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

Tag backscattering analysis,breast cancer,high frequency ultrasonic imaging,OAI-PMH Harvest,ultrasound-guided breast biopsy

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Shung, K. Kirk (committee chair), Chen, Mike S. W. (committee member), Yen, Jesse T. (committee member)

Creator Email akiyama@2002.jukuin.keio.ac.jp,takiyama@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-231707

Unique identifier UC11278909

Identifier etd-AkiyamaTak-4292.pdf (filename),usctheses-c40-231707 (legacy record id)

Legacy Identifier etd-AkiyamaTak-4292.pdf

Dmrecord 231707

Document Type Thesis

Format application/pdf (imt)

Rights Akiyama, Takahiro

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA