Research
article
Photoacoustic-guided
focused
ultrasound
(PAFUSion)
for
identifying
reflection
artifacts
in
photoacoustic
imaging
Mithun
Kuniyil
Ajith
Singh
*
,
Wiendelt
Steenbergen
BiomedicalPhotonicImagingGroup,MIRAinstituteforBiomedicalTechnologyandTechnicalMedicine,UniversityofTwente,P.O.Box217,7500AEEnschede,
TheNetherlands
1. Introduction
Photoacoustic(PA)imagingisapromisingbiomedicalimaging modalitythathasemergedoverthelast decade.InPAimaging, pulsed light absorbed by the target emits thermo-elastically generated ultrasound (US).Thisdata can bedetected using US probesallowingthereconstructionoflocationandspatialdetailsof thelight-absorbing target [1,2]. PA imaging thus combines the advantagesofUSandopticalimaging,providingexcellentoptical contrastwithultrasonicresolution.WhileUSimagingmakesuseof acousticscatteringandreflection intissuetoprovidestructural details, PA imaging extracts functional information based on opticalabsorptionbytissuechromophoressuchasblood.SincePA imaginginvolvesUSdetection,itcanberealizedinacommercially availableUS scannertoperform dualmode PA/USimaging [3– 5].Thesedualmodesystemspreferablyutilizeaschemeinwhich tissue is irradiated from the same side where PA signals are detected(reflection-mode,epi-modePAimaging)[6].Thismodein whichopticalcomponentsandUStransducersarecombined,aids thecliniciantoperformsinglehandguidanceoftheprobeduring
imaging.Inaddition,epi-illuminationmodefacilitatestheimaging ofbodypartswherebonesoracousticallyattenuatingsofttissue wouldobstructpropagationofacousticwavesfromthe illuminat-edtissueregiontotheacousticprobe.
Inmanyofthereportedhandheldprobe-basedPA/USsystems, light illuminationforPAimaging isdoneatanobliqueanglein suchawaythatitcoincideswiththeUSimagingplane[3–5]with thegoaltomaximizefluenceandthussignal-to-noiseratio.Onthe downsidethisresultsinahighlightfluenceonthesurfaceofthe tissuejustbeneaththeUSprobe,suchthatmelaninandsuperficial bloodvesselsgeneratestrongPAtransientswhichpropagateinto thetissueandreflectbackfromacousticallydensestructures[7– 9].Thesereflectedsignalsappearasartifactsinthereconstructed PAimages.Thereflectionartifactssignificantlyreducethecontrast, and thus theimaging depth,which is critical [7,8]. Sincethese artifactsaretriggeredbythepropertiesofthetissue,simplesignal averagingisnoteffectiveforreducingthem.Becauseofthestrong opticalattenuationintissue[1],reflectionartifactsthatshowupin acertaindepthcanbecomestrongerthanthePAsignalsofinterest inspiteofthelowlevelofacousticscattering[10],whichcanlimit theimagingdepth.Asanexample,whenimagingstructureslike finger joints (multiple light absorbers and acoustic reflectors), signalsofinterestmaygetmixedwiththereflectionsfromboneor tendon,which resultsinthewronginterpretation ofimages.To
A R T I C L E I N F O
Articlehistory:
Received12May2015
Receivedinrevisedform22August2015
Accepted23September2015
Availableonline28September2015
Keywords: Photoacoustic Epi-photoacoustic Ultrasound Imagingdepth Contrast A B S T R A C T
Influenceofacousticinhomogeneities andresultingreflectionartifactsis animportantproblemin reflection-modephotoacousticimaging.Absorptionoflightbyskinandsuperficialopticalabsorberswill generatephotoacoustictransients, whichtraverseintothetissueandgetreflected fromstructures havingdifferentacousticimpedance.Thesereflectedphotoacousticsignals,whenreconstructed,may appearintheregionofinterest,whichcausesdifficultiesinimageinterpretation.Weproposeanovel method to identify and potentially eliminate reflection artifacts in photoacoustic images using photoacoustic-guidedfocusedultrasound[PAFUSion].Ourmethodusesfocusedultrasoundpulsesto mimicthewavefieldproducedbyphotoacousticsourcesandthusprovidesawaytoidentifyreflection artifactsinclinicalcombinedphotoacousticandpulse-echoultrasound.Simulationandphantomresults arepresentedtodemonstratethevalidityandimpactofthismethod.ResultsshowthatPAFUSioncan identifyreflectionsinphotoacousticimagesandthusenvisagespotentialforimprovingphotoacoustic imagingofacousticallyinhomogeneoustissue.
ß2015TheAuthors.PublishedbyElsevierGmbH.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
* Corresponding author.Tel.:+31534892012;fax:+31534891105.
E-mailaddress:m.kuniyilajithsingh@utwente.nl(M.KuniyilAjithSingh).
ContentslistsavailableatScienceDirect
Photoacoustics
j ou rna l h ome p a ge : w ww . e l se v i e r. co m/ l oc a te / p a cs
http://dx.doi.org/10.1016/j.pacs.2015.09.001
2213-5979/ß2015TheAuthors.PublishedbyElsevierGmbH.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/
probe.Althoughthistechniqueshowedpromisingresults,notable disadvantageshavebeenreported:1)controlledprobemotioncan beperformedonlybyanexperiencedpersonandthistechnique canbe employed onlyfor easily deformable tissue, and 2) the maximumachievabletissuedeformationononesidelimitedby thetissuemechanicalproperties,andtheminimumdeformation requiredforartifactdecorrelationontheotherside,determinesthe contrast improvement. Another technique employs localized vibrationtagging[LOVIT]oftissueusingacousticradiationforce [ARF] for reducing clutter in the focal region of a long-pulsed ultrasonicbeam[7].Whileclaimingnearlyfullartifactelimination, authorsalso reported thedifficulty in eliminating echo clutter completely using LOVIT. For successful echo clutter reduction, LOVIT prefers a small ARF displacement region, which sets a limitationtothereal-timecapabilitybecauseextensivescanningis thenrequiredtoachieveclutterreductioninalargefield-of-view. This method also requires transducers that are capable of transmittingARFpulsesandislimitedbytheUSsafetyregulations. Inthispaper,weinvestigateanovelmethodthatcanidentify and potentially eliminate reflection artifacts in PA images. We specificallyaimatreducingthereflectionartifactscausedbyPA sourceswithin theimaging plane.Our technique ultrasonically simulatesthePAsignalfromanopticalabsorberanduncoversPA signalreflectionscausedbyacousticreflectorsbeneathoraround the absorber. The described method does not require any additionaltransducers orcomputationally intense algorithm to identify reflection artifacts, thus foresees good potential in improving real-time clinical PA imaging. Compared to LOVIT, ourmethodworkswithlowultrasoundpowerandthusdoesnot poseanyrisksinclinicalapplication.Becausethefocused pulse-echoacquisitionsinPAFUSioncanbeperformedatamuchhigher frameratethanPAimaging,itholdspotentialtobefastand real-time capable. Compared to DCA, it is not limited by tissue
Fig.1aillustratesthegenerationofaPAsignalfromanoptical absorber, and the resulting PA reflection signal caused by an acousticreflectordeeperinsidethetissue.InPAFUSion(Fig.1b),we transmitafocusedUSpulse,withthefocalpositionadjustedtothe locationoftheopticalabsorberbyusingguidanceofPAdata.The focusedUSpulse,atthetimeofarrivalinthefocalposition,mimics thepartofthePAwavethatwastraversingtowardsthereflector. Under theassumptionthatsignalacquisition startsatthetime when thefocusedUS pulsearrivesat thePAsource(t=0), the resulting signal willshow theUS reflections at thesame time wheretheyshowupinthePAsignal.WhenreconstructingaPA imagefromtheUSdata,itwillthusmimicthereflectionartifactsin termsofshapeanddepth withoutcontainingtherealPAsignal itself.
Thus, instead of identifyingreflectors, which would bedone withnormal US imaging, withPAFUSion we identifyPA signal reflections, which then can be used to correct PA images for reflectionartifacts.
3. Methodsandmaterials 3.1. PAFUSion–Processingsteps
TheprocessingstepsofPAFUSionareschematicallyillustrated in Fig. 2. Let us consider a simple medium with two optical absorbersandoneacousticreflectorinbetweenthem.Thedistance betweenopticalabsorber1andtheacousticreflectorisdasshown in Fig. 2a. Once after collecting PA data, an image can be reconstructedusinganyreconstructionalgorithm.Fig.2bshows theschematicofthereconstructedPAimage,inwhichthreehigh intensitypointsP1,P2,andP3arevisible.PointsP1,P2,andP3have
intensitiesI1,I2,andI3respectively.Intensityforaparticularpoint
Fig.1.(a)Illustrationofphotoacousticsignalgenerationandsignalcausingreflectionartifacts(b)PrincipleofPAFUSion:Focusingultrasoundtothephotoacousticsource
isdefinedhereasthemaximumvalueofthereconstructedimage envelopeatthelocationofthatpoint.ThedistancebetweenP1and
P2is2d.ThisPAimageisusedastheguidanceforfurtherstepsin
thetechnique.PAFUSionisappliedonpointsfromtoptobottomin PAimage.
IntheexampleinFig.2threehighintensitypointsareidentified inthePAimage,thustwostepsarerequiredtoperformPAFUSion. ItisnotnecessarytoapplyPAFUSiononP3,asthisdeepestfeature
inthePAimagewillneverleadtotheidentificationofagainanew reflection.Inthefirststep,USisfocusedontoP1(Fig.2c)andthe
resultingechoesareacquired.Reconstructionfromthisechodata isperformedconsideringone-waypropagationofsoundandby settingtheacquisitionstart(t=0)tothetimeatwhichUSreaches thefocusdistance.Bythisway,reconstructiontreatsthespatial pressuredistributionofthefocusedUSpulseastheinitialpressure distributionofavirtualPAsource,andthuswillbeabletomimic the reflection artifacts caused by that PA source. The image obtainedbythisstepissavedforfurtherprocessing.Inthesecond step,thesameprocedureisrepeatedforP2(Fig.2d).Sincethereis
noacousticreflectorbeneathP2inthisexample,resultantimageof
thesecondstepwillbeblankwithoutanyechoes.Duringeachstep, US reflections from thedepth of focus as wellas above it are omittedfromthereconstructedimage.Thenextstepisweighted additionofimagesobtainedinstep1andstep2forobtainingthe PAFUSion image (Fig. 2e), which reveals only the reflection artifacts. The weights for the images obtained in steps 1 and 2 are chosen to be proportional to the intensities I1 and I2
respectivelyinthePAimageinFig.2a.Asafinalstep,weenvisage thatthePAFUSionimagecanbeusedtocorrectthePAimagefor
obtainingareflectionartifact-freePAimage(Fig.2f).However,this proof-of-principlestudyfocusesonlyonidentifyingthereflection artifactsinPAimages.CorrectingthePAimagesusingPAFUSion imageswillberealizedasanextstageofimprovement.
3.2. Equipmentandsetup
A handheld dual-mode PA/US system which was already reported by ourgroup [5] wasusedfor all theexperiments in thisstudy.Fig.3showsthephotographofthesysteminwhicha commercial US scanner (MyLabOne, Esaote Europe BV, The Netherlands)wasusedalongwithaprobethatintegratesanUS array with a diode laser module emitting pulses at 805nm wavelength,130nspulsewidthandpulseenergyof0.56mJ.The US probehasa -6dBfractionalbandwidthofaround100% and centerfrequencyof7.5MHz.
ThesystemwasusedinresearchmodewhereUStransmission, laser pulse transmission, and data acquisition were controlled usingcustom-madesoftwarerunningonaPC.Thesamesoftware alsocontrolledtheswitchingbetweenPAimagingandUSfocusing forPAFUSiontechnique.Inaddition,plane-waveUSimageswere acquiredasareferencetomonitorreflectorpositions.USfocusing wasachievedbyadjustingthetransmissiondelaysofthedifferent transducerelementsinthelineararray.Bythisway,itispossibleto scan thefocus to any point in the imaging plane. A one-cycle transmissionpulseshapewasappliedtoeachtransducerelements. In plane-waveUS,PA,and PAFUSionimaging, RFdataofallUS transducer elements were saved after being acquired by the scanner with50MHz sampling frequency and digitized witha
Fig.3.Portableultrasoundscanner(left)andthehybridprobe(right)integratinglasermoduleandUStransducerarray[5].
Fig.2.Step-by-stepschematicillustrationofdataprocessingstepsinPAFUSion.(a)Simplemediumwithtwoopticalabsorbersandoneacousticreflector,(b)Photoacoustic
image,(c)PAFUSionimagingstep1–USfocusonfirsthigh-intensitypointinthePAimage,(d)PAFUSionimagingstep2–USfocusonsecondhigh-intensitypointinthePA
imagingandreconstruction2)focusedUSimaging1(focustofirst highintensitypointinthePAimage),and3)focusedUSimaging2 (focusto secondhigh intensitypoint in the PAimage). Finally weighted addition of images (weight proportional to the PA intensityat thedepthsof focus)obtained in step 2 and 3 was performed forobtainingthe PAFUSionimage. For allthe steps, reconstructionwasdone using a 2-D frequency domain recon-structionalgorithm[14].
3.4. Phantommeasurements
Threephantommeasurementswereperformedforprovingthe validityofthePAFUSiontechniquetoidentifyreflectionartifactsin PAimaging.Beforeallexperiments,measurementsweredoneby movingtheUS/PAprobeforthandbackaxiallytoidentifypotential reflectionartifactscausedbyopticalabsorptiononthetransducer surfacethatwouldmoverelativetothePAsignal.Inthatwaywe madesurethatallthereflectionartifactsarereallycausedbyPA sourcesinsidethephantom.Thefirstphantomwassimilartothe digital phantom in the simulation study. Fig. 5a shows the schematicofthisphantominwhichoneopticalabsorber(nylon thread) and acoustic reflector (delrin rod) was used. Both the opticalabsorberandacousticreflector werepositioned inside a tankfilledwithwaterinsuchawaythattheywereperpendicular totheimagingplaneoftheUS/PAprobethatwasimmersedin water(Fig.5a).Thisphantomwasusedtostudyasimplesituation withasinglephotoacousticreflectionanditsidentificationusing
problemin imagingstructureslikefinger joints,where multiple absorbersandacoustic reflectorsarepresent.Water mixedwith Intralipid(ms’=6cm-1)wasusedasthemediuminthisphantomto
makesurethatbothnylonthreadsgeneratedasignificantPAsignal. Thenylonthreadthatwasusedas theopticalabsorberwas itselfacousticallyreflective,thusitcancausereflectionartifacts even in theabsence of additional acousticreflectors. Thethird phantom (Fig. 7a) represents the condition in which optical absorbersthemselvesareacousticallyreflective,andwasmeantto study and identify theresulting reflection artifacts.Two nylon threads were kept at different depths in water and were perpendicular totheUS/PA probe (Fig. 7a).Water withoutany scatteringwaschosenasthemediuminthisphantomforobtaining astrongPAsignalfromthefirstnylonthread(andtherebyastrong reflectioninterferingwiththesecondnylonthread)bydirecting thelighttothetoppartofthephantom.
4. Results
Inalltheresults,lateralandaxialcoordinatesarerepresented by x and z respectively and the envelope of the images (PA, PAFUSion and US) areplotted in linear amplitude scale. Image reconstructionassumedconstantspeedofsoundinthemedium forallphantommeasurements,whichmayhaveresultedinaslight deviationofthedepthofreconstructedfeaturescomparedtothe physicaldepthowingtotheinhomogeneousspeedofsoundofthe embeddedinclusions.
Fig.4.(a)Digitalphantomusedforsimulation,includingtheacousticproperties,(b)photoacousticimageandenlargedregionofinterest,whichshowsthereverberation-like
4.1. PAFUSionsimulation
Fig.4a shows the details aboutthe phantom and US probe positioningusedforthePAFUSionsimulation.Fig.4bshowsthe reconstructedPAimage.ThePAsource(x=8mm,z=4.5mm)and itsreflection on the acousticreflector which occursat position (x=8mm,z=9.5mm)areevidentinthePAimage.Limitedview artifactsintheshapeofcircularstreaksarevisibleinbothPAimage and PAFUSion images. The PAFUSion image (Fig. 4c) clearly identifies the PA reflection artifact by reproducing the correct depthandapproximatelythecorrectshape.Itisevidentfromthe resultsthat PAFUSioncouldreproduce eventheminutelayered artifacts seen below the high intensity reflection (x=8mm, z=9.5mm). The intensity of the artifacts is different in the PAFUSionimagethaninthePAimagebecausethesimulatedUS transmissionpressureamplitude waschosenindependent from thePAsignalamplitude.
4.2. Phantomexperiments
Fig. 5a shows the details about the first phantom and the schematicofthemeasurementsetup.Thisphantomhadonlyone opticalabsorberembeddedandanartifact-freePAimagewould haveonlyonehighintensityspot.Fig.5bshowsthereconstructed PAimagein whichthesignalfromthenylonthread(x=8mm, z=4.5mm) and two other signals (x=8mm, z=7mm and x=8mm, z=9.5mm) are visible. The nylon thread itself (x=8mm,z=4.5mm)andtwosurfacesofdelrinrod(x=8mm, z=5.9mmandx=8mm,z=7.3mm)areobservableinthe plane-waveUSimage(Fig.5c).ThePAFUSionimage(Fig.5d)evidently revealed two reflection signals and confirmedthat signals at x =8mm,z=7mmandx=8mm,z=9.5mmin thePAimageare reflectionartifacts.Itisworthmentioningthatreflectionartifacts identifiedbyPAFUSionreproducedtheshapeandspatialdetailsof theactualPAreflectionartifactsquitewell.Thesecondreflection
Fig.6.(a)Schemeofphantom2andexperimentalsetup,(b)photoacousticimage,(c)plane-waveultrasoundimage,(d)PAFUSionimage.
artifact(x=8mm,z=9.5mm)iscausedbytheechoofthenylon thread PA signal on the second surface of the delrin rod. An extendedhorizontalfeaturecanbeseenrighttothefirstidentified reflection artifact (x=8mm, z=7mm)in the PAFUSion image (Fig.5d).Thisispotentiallyareconstructionartifactcommontothe frequency-domain algorithm, which is better visible in the PAFUSionimagethaninthePAimageowingtoahighercenter frequencyofthetransmittedUScomparedtothePAsignal.These artifactsoccur outside theregion of interest and are thus not criticalforthisexperiment.Theintensityratioofthetwoidentified reflections (x=8mm, z=7mm and x=8, z=9.5mm) in the PAFUSionimageideallywouldbethesameastheratioofreflection intensitiesinthePAimage.However,thisratiowasfoundtobe differentinthePAFUSionandPAexperiments(10timeshigherin PAFUSionimages).Theintensitiesofthesecondreflectionsappear similarinthePAimageandthePAFUSionimage,becausethecolor map in the PAFUSion image was chosen such that the pixel intensityofthefirstartifactwassaturatedtoportraybothartifacts clearly.
Fig.6a showsthedetailsaboutthesecondphantomand the schematicof the arrangement used for themeasurement. This phantom consisted of two optical absorbers and an acoustic reflector.Ideally,onewouldexpectonlytwohighintensityspotsin thePAimage.Fig.6bshowsthereconstructedPAimageinwhich thesignalsfromthetwonylonthreads(x=8mm,z=5mmand x=9.065mm,z=10.4mm)andtwoothersignals(x=8,z=7.4and x=8,z=9.8)arevisible.Bothnylonthreads(x=8mm,z=5mm and x=9.065, z=10.4mm) and two surfaces of delrin rod (x=8mm, z=6.3mm and x=8, z=7.7mm) are noticeable in theplane-waveUSimage(Fig.6c).ThePAFUSionimage(Fig.6d) clearlyrevealedtworeflectionsignalsandconfirmedthatsignals atx=8mm,z=7.4mmandx=8mm,z=9.8mminPAimageare reflection artifacts. The second reflection artifact (x=8mm, z=9.8mm)isalmostfusedwiththesignalfromthesecondnylon thread(x=9.065mm,z=10.4mm),mimickingaclinicalscenario where identification of such an artifact would be crucial. The extended horizontal feature right to the first reflection in the PAFUSionimageisagainvisible.Thisfeatureoccursoutside the regionofinterestandthusdoesnotpreventtheidentificationof
reflectionartifacts.Thedistancebetweenthefirstnylonthreadand thedelrinrodinthisphantomisthesameasforthefirstphantom. Alsotheintensityratiosofthetwoidentifiedreflections(x=8mm, z=7.4mmand x=8mm,z=9.8mm)weresimilar totheones found in phantom 1. This is reasonable because the distance travelled by the US (PAFUSion, PA imaging) and thus the attenuationwassimilarinphantom1andphantom2.
Thethirdphantomwasdesignedtosimulatetheconditionof opticalabsorbersthatthemselvesreflectPAsignalsfromotherPA sources. Recent studies shown that in PA finger imaging, the tendonshowscontrastinPAaswellasUSimaging[16].Iftheskin/ blood vesselsignal gets reflected on the tendon, the resulting artifactscanmakeimageinterpretationdifficult.
Fig. 7a shows the details of the third phantom and the schematic of the measurement setup. Fig. 7b shows the reconstructed PA image(zoomed in for moredetails)in which signals from the two nylon threads (x=8mm, z=4mm and x=8mm,z=6.2mm)andanothersignal(x=8mm,z=8.5mm) areevident.Bothnylonthreadsarevisibleintheplane-waveUS image(Fig.7c).ThePAFUSionimage(Fig.7d)clearlyexposesthe reflection signal and confirms that the feature at x=8mm, z=8.5mm in the PA image is a reflection artifact. Itis worth mentioning thatthePAFUSionimagereproducesthedepthand shapeofthereflectionartifact.
5. Discussion
Forthesimulationandallphantommeasurements,ourresults showthatPAFUSion is capableof identifyingall thereflection artifactspresentinthePAimage.Ourphantommeasurementsled topromisingresultsforstructuresseparatedbyaround1.5mm, whichisquiteclosetotheclinicallyrelevantscenarioinfinger jointimaging(distancebetweenabloodvesselandatendon,see furtherbelow).Astraightforwardmethodtoidentifyreflection artifactswouldbetousetheB-modeUSimageandtoidentify potentialacousticreflectorsinthem,andthenusethis informa-tiontogether withsimulationstoidentifyreflectionsin thePA image. However, the US simulations would then rely on an imperfectinputdataset,andfurthermorethetypeofultrasound
Fig.7.(a)Schemeofphantom3andexperimentalsetup,(b)photoacousticimageandenlargedregionofinterest,whichshowsthereverberation-likedetailsofthereflection
acquisitionusedintheUSimage(centerfrequency,numberof cycles,transmissionangle)isoftendifferentfromthetypeusedin thePAimage.Furthermore,thecomputationalcostofawave-field simulation would limit the real-time applicability of such an approach.Forthesereasons,itcanbeadvantageoustophysically mimic the reflection artifacts instead, by using ultrasound transmissionsthatmatchthewavefieldofthePAsourcesina physical back-propagation approach. The key aspect in the PAFUSionprocessisthatthezerotimehastobedefinedatthe moment whentheUSpulse reachesthe absorber,rather than whenthepulseisinjectedinthetissueasinnormalUSimaging. Hence,inthePAFUSionalgorithm,thetissueitselfrevealsthePA reflectionsbyapplyingUSimaginginamannerthatsimulatesthe timingof PA images. Another critical featurein thePAFUSion processisthattheUSpressuredistributionmatchesthePAinitial pressuredistribution.IftheshapesofUSandPAinitialpressure distributionsarenotmatched,itwillresultinwrongarrivaltime ofthe transientat differentreflectors,becausethewavefront curvatureofthedivergingUSwavefrontwillbedifferentfromthe PAwavefront.
AnydualmodePA/USsystemcanincorporatethePAFUSion technique without any special changes to the system. The software requirements are quite similar to those of normal line-by-lineUSimaging, whichmakes clinicalimplementation straightforward. There is no training required to use this techniquesinceeverythingcanbesoftwarecontrolledjustasin USimaging.Inthiswork,thePAFUSionimagewasobtainedby weightedadditionofimagesobtainedindifferentsteps,inwhich theweightisproportionaltothePAsignalintensity.Inavariation onthisprocedure,theUSpulseamplitudecanalsobevariedbased onthePAsignalintensity.Thiswillreducetheoverheadofdoing weightedadditionofimagesduringprocessing.Anotherpractical limitationisthatPAFUSionrequiresanextrameasurementand computationstep,whichmaymakethetotalimagingprocedure slower.However,weexpectthatdevelopmentsinGPU(Graphical Processing Unit) and FPGA (Field Programmable Gate Array) technology can overcome this, as parallel acquisition and processingisfeasibleusingthesehigh-speedtechniques.Speed ofsoundandacousticattenuationvariationsintissuemayalso playamajorroleinaccuracyofthistechniquewhenappliedin vivo.Futureworkwillalsofocusonconsideringtheseparameters duringprocessing.
Thisworkwasintendedforpresentingtheproof-of-principleof the technique and therefore concentrated on a comparably simplified scenario where the reflection artifacts were present in thecenter partof theUS imaging plane, and PAFUSion was applied on PA images containing concentrated features (nylon thread).However,thechoiceofthephantomsisnotalimitationto thevalidityofthetechnique.Featureswithamorecomplicated shape,andfeaturesthatarespatiallyextendedareexpectedinan invivoscenario.Weenvisagesolvingtheseinourfuturestudiesby transmittingUSpulseswiththeshapeoftheidentifiedPAfeature, and then applying PAFUSion algorithm to identify reflection artifacts.
Thisworktargetedonlyatidentifyingthereflectionartifacts, but not at their elimination. Elimination of artifacts requires furtherinvestigationintoathoroughcalibrationofpulseshapeand amplitudeof thefocused US transmissions. Such an effort was beyondthegoalofthisstudy,butwillberealizedasanextstepof improvement. Two critical aspects to consider for making the reflection-artifacteliminationworkinournextstepare:
ShapeofUSfocus:ThecharacteristicsoftheUSfocushavean impact on the shape and intensity of identified reflection artifactsinPAFUSion.Itiscriticaltohaveanarrowfocusaxially andlaterallytomimicsmallPAfeatureslikebloodvessels.Using aSchlieren-imagingsetup,wecharacterizedthesizeandshape
of the USfocus.Whenfocusing USto adepthof 4.5mm,we achievedafocuswithlengthof0.2mmandwidth0.4mminthe imagingplane.Thisshowsthat,usingoursystem,wewillbeable toapplyPAFUSiononsmallbloodvessels.Intherangeofdepth thatwetarget(until15mm),changeinshapeofthefocuswith respecttothedepthwasfoundtobenegligible.
Frequencycontent:ThepulseshapeofthetransmittedUSin PAFUSionmustbeideallythesameasthePApulseshapetobe able to coherently subtract PA reflections and PAFUSion-identified reflections. Inthis study, the frequencycontentof theUSfocusedataPAfeatureinthePAFUSionprocedureisnot matchedwiththefrequencycontentofthePAsignalscoming fromthesefeatures.Consideringthepulsewidthoflaserdiodein ourintegratedprobe(130ns),theremaybealowpassfiltering effect.Thus,thereflectionsinPAimagesareoflowerfrequency (4MHz),partlyoutsidethebandwidthofthetransmittedUSin PAFUSionmeasurements.Thelowfrequencyof thePAsignals might be oneof the reasonsfor the reflection-intensityratio difference seen in the PA image and PAFUSion image in the phantomexperiments.Matchingthefrequencycontentwillbe onefocus of futureinvestigation, and canbeachievedeither physically by adapting the US transmission spectrum, or in softwarebyfiltering.
At this point, we would like to drawattention tothe side-constraintsandfurtherstepsthatwillbeimportantwhenworking towardsaclinicalimplementationoftheproposedtechnique.
First of all, we remind the reader thatPAFUSion can only identify and compensate for reflection artifacts caused by PA signalsgeneratedinsidetheimagingplane,butnotout-of-plane clutter. This is not a strong limitation, because PAFUSion is specifically designedfor a setupwherethetissueis irradiated directlybelowtheprobetomaximizeSNR.Normally,theresulting strongreflectionartifactsinhibittheuseofsuchasetup,andthe optimum irradiation distance is determined by a trade-off between in-plane reflection artifacts and out-of-plane clutter and SNR [12]. If PAFUSion manages to reduce the in-plane reflection artifacts, irradiation directly below the probe will become anoption again,andthen out-of-planeclutterwill be insignificant.
Second, PAFUSioncanonlyidentifyandcompensate forthe inwardpropagatingpartofPAsignalsthatcanbemimickedusing thelimitedrangeofpossibleUStransmission angles.Atypical ultrasoundprobehasanangularapertureintherangeof-308to 308,whereastypicalPAsources(cylindricalbloodvessels)radiate into ananglerangeof 3608,whichleadstoalimitation ofthe method. If an acoustic reflector and optical absorber are positioned sideby side, thenthe reflection artifact caused by these will be impossible to identify using this technique. Consideringthefactthatmostoftheechoproducingstructures (tendon,bone)liebeneaththesuperficialopticalabsorbersthat generatestrongPAtransients(skin,bloodvessels),thislimitation isnotcritical.
Also,whenusingalineararrayprobe,PAFUSionislimitedto mimicking PAtransients thatpropagateparallel totheimaging plane.Therefore,theprobe mustbeorientedsuchthat artifact-generatingPAsourcesareorientedperpendiculartotheimaging planeinordertoobtainoptimumperformance.Thiscanbeeasily achievedinfree-handprobeguidance,andwillbeevenlessofa limitationif1.5Dor2Darraysareused.
WithincreasingnumberofPAabsorbers(skin,multipleblood vessels) and many possible reflection angles (angle between absorber to bone/tendons), identification of all the reflection artifactsmaybechallenging.However,itisvitaltomentionthatan artifactreductionby75%facilitatesanincreaseinsignaltoartifact ratioby12dB.ClinicalPAimagingsuffersfrompoordeep-tissue contrastbecauseofopticalattenuationandreflectionartifacts,and
UScanbesyntheticallyfocusedtoanydesiredpointintheimaging plane.Insteadof physicallyscanninge.g.100100pixelswitha focused beam, the focused acquisitions can be synthetically generatedinsoftwareprocessingfrome.g.just 100plane-wave transmissionsusingdifferenttransmitangles.
Time reversal of photoacoustic signals could also be an alternative for reducing acquisition time in comparison to separatelyfocusingonmultiplePAsources.Timereversalcould beasolutionfor twoproblems:(1)thePAimages oftenshow spread-outfeaturesratherthanpoint-likefeatures,and(2)the differentspectralcontentsofPAandUSsignals.Timereversalmay leadtoingoingUSsignalswhicharespectralcopiesofoutgoingPA signals.Theonlyproblemtowhichtimereversalisnotasolution isthatoffrequency-dependentattenuation.Ourfutureresearch willthereforefocusonstudyingthefeasibilityofusingthePAtime reversal approach in PAFUSion. Most of the commercially availableUSsystemsmaynotbecapableofpulsingthetransducer elementswithnon-periodicpulses,whichwouldbethetechnical challengeinusingPAtimereversal.Thislimitationcanalsobe solvedbyusingasyntheticapproachwheretheUStransmission pulseshapeismatchedwiththePAsignalshapeusingsoftware processing.
AtthisstageofPAFUSionimplementation,oneoftheclinical applicationsweforeseeisfingerjointPAimaging.Forinstance,if thedistancebetweenasuperficialbloodvesselandatendonisthe sameasthedistancebetweenthetendonandsynovium,theblood vesselsignalreflectsonthetendontogeneratereflectionartifact atalmostthesamedepthasthesynoviumislocated.Thisiscritical inrheumatoidarthritisimagingwhereinflammationofsynovium isthemarker.Thesecondphantomexperimentwasintendedto mimicthissituationanditisclearthatPAFUSioncanbeusefulin thesecircumstances.Thethirdphantomexperimentrepresents the condition in which optical absorbers are by themselves acousticallyreflective.ThesefeatureswithcontrastinPAandUS imaging might create problems in accurate interpretation of clinicalPAimages.Oneoftheimportantreflectionartifactsources istheskinbecauseofits highmelanincontent.Inanyclinical applications, skin is expected to generate high PA transients, whichtriggersthegenerationofreflectionartifacts.PAFUSionwill helptoavoidtheexclusionofpatientswithhighmelanincontent, fromdeep-tissuePAimaging. Reflectionartifactsgenerated by smallmolesorhair,whichisdirectlyundertheprobecanalsobe potentiallyidentifiedbyusingPAFUSion.
6. Conclusions
PAFUSion allows the identification of reflection artifacts in photoacousticimagesbyultrasonicallysimulatingthePAwaves from the optical absorber, traversing towards the acoustic reflectorsandthusbymimickingthePAreflectionsignals.Inthis
Acknowledgement
Research was funded by the European Community’s Seventh FrameworkProgramme(FP7/2007-2013)undergrantagreementn8 318067. Authors acknowledge Dr.M. Jaeger,Institute ofApplied Physics,UniversityofBernforprovidingsoftwareforthe Fourier reconstructionalgorithmandforhisvaluablecommentsonourwork. AuthorsalsoacknowledgeKhalidDaoudi,PimvandenBerg,Altaf Hussain,andJacobStaleyfortheirinsightfuldiscussionsandhelp. References
[1]M.Xu,L.V.Wang,Photoacousticimaginginbiomedicine,ReviewofScientific Instruments(2006)77,41101-1–22.
[2]P.Beard,Biomedicalphotoacousticimaging,InterfaceFocus1.4(2011)602–631.
[3]R.G.M.Kolkman,P.J.Brands,W.Steenbergen,T.G.vanLeeuwen,Real-timein vivophotoacousticandultrasoundimaging,JournalofBiomedicalOptics (2008),13.050510-1-3.
[4]C.Kim,T.N.Erpelding,L.Jankovic,L.V.Wang,Performancebenchmarksofan array-basedhand-heldphotoacousticprobeadaptedfromaclinicalultrasound systemfornon-invasivesentinellymphnodeimaging,Philosophical TransactionsoftheRoyalSocietyA:Mathematical,PhysicalandEngineering Sciences369.1955(2011)4644–4650.
[5]K.Daoudi,P.J.vandenBerg,O.Rabot,A.Kohl,S.Tisserand,P.J.Brands,etal., Handheldprobeintegratinglaserdiodeandultrasoundtransducerarrayfor ultrasound/photoacousticdualmodalityimaging,OpticsExpress22.21(2014) 26365–26374.
[6]M.KuniyilAjithSingh,W.Steenbergen,S.Manohar,HandheldProbe-Based DualModeUltrasound/PhotoacousticsforBiomedicalImaging,in:M.Olivo, U.S.Dinish(Eds.),FrontiersinBiophotonicsforTranslationalMedicine,vol.3, Springer,Singapore,2016,pp.209–247.
[7]M.Jaeger,J.C.Bamber,M.Frenz,Cluttereliminationfordeepclinical optoacousticimagingusinglocalisedvibrationtagging(LOVIT),Photoacoustics 1.2(2013)19–29.
[8]M.Jaeger,L.Siegenthaler,M.Kitz,M.Frenz,Reductionofbackgroundin optoacousticimagesequencesobtainedundertissuedeformation,Journalof BiomedicalOptics(2009)14,054011-1-10.
[9]M.Frenz,M.Jaeger,Optimizationoftissueirradiationinoptoacousticimaging usingalineartransducer:theoryandexperiments,PhotonsPlusUltrasound ImagingandSensing(2008)6856,68561Y-1-13.
[10]M.Jaeger,M.Frenz,D.Schweizer,Iterativereconstructionalgorithmfor reductionofechobackgroundinoptoacousticimages,PhotonsPlus UltrasoundImagingandSensing(2008)6856,68561C-1-15.
[11]M.Jaeger,K.Gashi,S.Peeters,G.Held,S.Preisser,M.Frenz,Real-timeclutter reductioninepi-optoacousticimagingofhumanvolunteers,MedicalImaging (2014)9040,90400N-1-5.
[12]G.Held,S.Preisser,H.G.Akarcay,S.Peeters,M.Frenz,M.Jaeger,Effectof irradiationdistanceonimagecontrastinepi-optoacousticimagingofhuman volunteers,BiomedicalOpticsExpress5.11(2014)3765–3780.
[13]M.A.L.Bell,N.Kuo,D.Y.Song,E.M.Boctor,Short-lagspatialcoherence beamformingofphotoacousticimagesforenhancedvisualizationofprostate brachytherapyseeds,BiomedicalOpticsExpress4.10(2013)1964–1977.
[14]M.Jaeger,S.Schupbach,A.Gertsch,M.Kitz,M.Frenz,Fourierreconstruction inoptoacousticimagingusingtruncatedregularizedinversek-space interpolation,InverseProblems23(2007)S51–S63.
[15]B.E.Treeby,B.T.Cox,k-Wave:MATLABtoolboxforthesimulationand reconstructionofphotoacousticwavefields,JournalofBiomedicalOptics (2010)15,021314-1-12.
[16]G.Xu,J.R.Rajian,G.Girish,M.J.Kaplan,J.B.Fowlkes,P.L.Carson,etal., Photoacousticandultrasounddual-modalityimagingofhumanperipheral joints,JournalofBiomedicalOptics(2013)18,010502-1-3.
Mithun Kuniyil Ajith Singh received a Technical
DiplomainMedicalElectronicsfromModelPolytechnic
College,Calicut,India,in2004,andaBachelordegreein
BiomedicalInstrumentationandEngineeringfromSRM
University,Chennai,India,in2008,andaMasterdegree
inMolecularNanoBio-photonicsfromE´colenormale
supe´rieure de Cachan, Paris, France, in 2012. Heis
currently working towards a PhD degree at the
BiomedicalPhotonicImaginggroup,MiraInstitutefor
BiomedicalTechnologyandTechnicalMedicine,
Univer-sityofTwente,Enschede,TheNetherlands.Hiscurrent
research focuses onthe development ofa clinically
feasible, low-cost, portable, multi-wavelength dual
modePhotoacoustic/Ultrasoundsystem.
WiendeltSteenbergenreceivedhis M.Sc.degreein
aerospaceengineeringin1988attheDelftUniversityof
Technology.HeobtainedaPhDdegreeattheEindhoven
UniversityofTechnologyin1995,specializinginoptical
measurementandmodelingofturbulentflow.Since
thenhehasbeenworkingattheUniversityofTwente,
from2000asassistantprofessor,from2008asassociate
professorandfrom2010asprofessorinbiomedical
optics.Hisresearchinterestsareincoherenttechniques
fortissuediagnosissuchaslowcoherence
interferom-etryandlaserspeckletechniquesforflowimaging,and
