Current Developments on Optical Feedback Interferometry as an All-Optical Sensor for Biomedical

27 Pages·2017·3.6 MB·English

Current Developments on Optical Feedback
Interferometry as an All-Optical Sensor for Biomedical
Applications
Julien Perchoux, Adam E Quotb, Reza E Atashkhooei, Francisco E Azcona,
Evelio Esteban Ramírez-Miquet, Olivier E Bernal, Ajit E Jha, Antonio Luna
Arriaga, Carlos E Yanez, Jesus E Caum, et al.
To cite this version:
Julien Perchoux, Adam E Quotb, Reza E Atashkhooei, Francisco E Azcona, Evelio Esteban Ramírez-
Miquet, et al.. Current Developments on Optical Feedback Interferometry as an All-Optical Sensor
for Biomedical Applications. Sensors, 2016, 16 (5), pp.694. ￿10.3390/s16050694￿. ￿hal-01389568￿
HAL Id: hal-01389568
https://hal.archives-ouvertes.fr/hal-01389568
Submitted on 28 Oct 2016
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est
archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
sensors
Article
Current Developments on Optical Feedback
Interferometry as an All-Optical Sensor for
Biomedical Applications
JulienPerchoux1,AdamQuotb1,RezaAtashkhooei2,†,FranciscoJ.Azcona2,†,
EvelioE.Ramírez-Miquet1,3,†,OlivierBernal1,AjitJha2,AntonioLuna-Arriaga1,
CarlosYanez2,JesusCaum2,ThierryBosch1andSantiagoRoyo2,∗
1 LAAS-CNRS,UniversitédeToulouse,CNRS,INP,6AlléeEmileMonso,31400Toulouse,France;
[email protected](J.P.);[email protected](A.Q.);[email protected](E.E.R.-M.);
[email protected](O.B.);[email protected](A.L.-A.);[email protected](T.B.)
2 CentrefortheDevelopmentofSensors,InstrumentsandSystems,UniversitatPolitècnicadeCatalunya
(UPC-CD6),RamblaSantNebridi10,E08222Terrassa,Spain;[email protected](R.A.);
[email protected](F.J.A.);[email protected](A.J.);[email protected](C.Y.);
[email protected](J.C.)
3 CentrodeAplicacionesTecnológicasyDesarrolloNuclear,Calle30,No.502,Miramar,
LaHabana11300,Cuba
* Correspondence:[email protected];Tel.:[phone]
† Theseauthorscontributedequallytothiswork.
AcademicEditor:VittorioM.N.Passaro
Received:1March2016;Accepted:6May2016;Published:13May2016
Abstract: Opticalfeedbackinterferometry(OFI)sensorsareexperiencingaconsistentincreasein
theirapplicationstobiosensingduetotheircontactlessnature,lowcostandcompactness,features
that fit very well with current biophotonics research and market trends. The present paper is a
review of the work in progress at UPC-CD6 and LAAS-CNRS related to the application of OFI
to different aspects of biosensing, both in vivo and ex vivo. This work is intended to present the
variety of opportunities and potential applications related to OFI that are available in the field.
Theactivitiespresentedaredividedintotwomainsensingstrategies: Themeasurementofoptical
path changes and the monitoring of flows, which correspond to sensing strategies linked to the
reconstruction of changes of amplitude from the interferometric signal, and to classical Doppler
frequencymeasurements, respectively. Foropticalpathchangemeasurements, measurementsof
transientpulses,usualinbiosensing,togetherwiththemeasurementoflargedisplacementsapplied
to designing palliative care instrumentation for Parkinson disease are discussed. Regarding the
Doppler-basedapproach,progressinflow-relatedsignalprocessingandapplicationsinreal-time
monitoringofnon-steadyflows,humanbloodflowmonitoringandOFIpressuremyographsensing
willbepresented. Inallcases,experimentalsetupsarediscussedandresultspresented,showingthe
versatilityofthetechnique.Thedescribedapplicationsshowthewidecapabilitiesinbiosensingofthe
OFIsensor,showingitasanenableroflow-cost,all-optical,highaccuracybiomedicalapplications.
Keywords: optical feedback interferometry; biosensors; vibrometry; flowmetry;
biophotonics;metrology
1. Introduction
Despitetheextensiveuseofsemiconductorlasers(SCLs)indiverseapplications,fromveryearly
times,itwasshownthattheymightshowrelevantinstabilitieswhensubjecttoexternalperturbations
as optical reinjection, also known as optical feedback (OF). When a fraction of the radiated field
Sensors2016,16,694;doi:10.3390/s16050694 www.mdpi.com/journal/sensors
Sensors2016,16, 694 2of26
from the laser is fed back into its own cavity, it induces a very rich phenomenology [1], including
modulation of injection current and emitted power, changes in emission modes and even chaotic
behavior,dependingontheamountofenergythatre-entersthecavity.Beyondthepotentialinstabilities
thatmaybeintroducedinthecavity,OFhasfoundseveralapplicationsinnon-destructivetesting
andnon-contactopticalsensingthroughtheopticalfeedbackinterferometer(OFI)(alsoknownas
self-mixinginterferometer(SMI))[2,3],wherethefieldemittedfromthelaserincidesuponaremote
target after traveling through an optical path, so changes in the few nanometers’ scale in such an
opticalpathbetweenthelaserandtargetmaybemonitored. Thebeatingofthedelayedopticalfield
fedbacktothecavitywiththatofthestationaryfieldinsidethecavitycausestheopticalfrequency
andpowerofthelasertochange,creatingobservableinterferencefringes,witheachfringeequivalent
toachangeofhalftheemissionwavelength[4]intheexternalopticalpathlengthseparatingthelaser
andthetarget. Suchachangecontainsthesignatureofrelevantpropertiesofchangesintheoptical
path,whichmaybemeasuredinthesamelaserusingthebuilt-inphotodiodeusedastheemission
monitor. The main advantage of OFI sensors is the use of the packaged laser as the light source,
theinterferingmediumandthecoherentdetectorallinone,makingthesetupextremelycompact,
economic,self-alignedandefficient,whilekeepingitcontactless. Further,theresolutionandaccuracy
obtainedwiththeuseofOFIiscomparabletothatofclassicalopticalinterferometry,asfarasitisa
coherentdetectionmethod[5].
Awidevarietyofapplicationshasbeenapproachedalongtheyearsindifferentfields,including
all types of changes in the optical path of the laser and Doppler effects in the cavity. Most of the
applications include the measurement of distance changes or the velocity of the target, but also
absolutedistance,profiling,measurementorpositioninghasbeenapproached[6–8]. Asthesignal
obtained may be quite noisy, accurate signal processing and extraction of the desired parameters
isrequiredfortheoptimumperformanceofOFIsensorsandhasbeenasourceofconstantactivity.
Differentapproacheshavebeentaken,includingsimplefringecounting(FC)[9],Fouriertransform
processing[10],thephaseunwrapmethod(PUM)[11]and,morerecently,theHilberttransform[12]or
wavelets [13], in order to extract the desired parameters (amplitude, velocity, distance, etc.) with
increasedaccuracy. Thishasledtoagrowingnumberofreal-worldapplicationsinsensing,asdiverse
asmotorrunout[14],integralstrain[15],laserparameters[16]oracousticfields[17],tonameafew.
Theinterestinthetechniquehasalsobeenacceptedindifferentindustrialenvironments,bothinthe
formofpatentapplicationsorasconcreteproductdevelopments[18,19].
Furthermore, although most of the activity on OFI has been traditionally happening in the
short and mid-infrared wavelengths, in the last few years, an entirely full field is emerging
related to terahertz self-mixing sensing, driven by the evaluation of novel quantum cascade
laser designs and properties. The research activities presented are combining the simplicity and
resolutioncapabilitiesofself-mixinginterferometrywiththeparticularopticalpropertiesofterahertz
radiation. Thisispushingforwardanumberofnovelapproachestosensors. ApplicationsinTHz
wavelengths are blooming, including large opportunities in the biomedical field [20], but also in
spectroscopy[21],materialanalysis[22]includingthedetectionofplasticexplosives[23],andimaging,
bothbyscanning[24]orbyusingasyntheticapertureapproach[25].
Biophotonics and, in particular, biomedical applications are experiencing, at the same time,
atrendtowardsbetter,lessexpensiveandmorecompactsystems,eitherfordiagnosticorprognostic
usesorinremotepatientcare. Suchatrendis,thus,extremelywellalignedwiththecapabilitiesofOFI
sensors. Althoughthetrendhasacceleratedinthelastfewyears,biophotonicshasbeenalwaysatopic
closetothedevelopmentsonOFI[26],whereapplicationsinpulseshapemeasurementsbasedon
Dopplermeasurements,bothinextracorporealcirculatorsorinvivo[27–29],havebeendemonstrated.
Optical path change measurements on the fingernail have also enabled the measurement of the
arterialpulsewave[30]. Othermeasurementsrelatedtobodyvibrationshavebeenperformedfor
respiratorymotility[31],orocularmovements[32],orevenheaddisplacement[33]. Otherapproaches
Sensors2016,16, 694 3of26
tothedevelopmentofinstrumentationhaveinvolvedthecombinationofOFIinterferometerswith
microscopy[34],includingconfocalarrangements[35].
Thispaperisintendedtoprovideageneraloverviewofthedifferentworkinprogressinour
labs,coveringthenoveldevelopmentsbeingrelatedtoOFItechniquesappliedtobiosensing. Wehave
dividedthemintothosebasedonsensingopticalpathchangesandthosebasedonDopplersensing.
Unlessexplicitlystatedotherwise,allsetupsareacquiringtheOFIsignalfromtheinternalphotodiode
ofthelaser,usingeitheraUSB-connectedoscilloscopeoranA/DconversioncardconnectedtoaPC.
OnlywhentheprocessingmethoddepartsfromthegeneralapproachinOFIdescribedin[2]will
beexplicitlydescribedforconciseness. Alongthenextsection,wewillintroduceourdevelopments
relatedtosensingofopticalpathchanges,appliedtothemeasurementoftransientvibrationsand
toanOFIsensorforverylargedisplacementstailoredtothedevelopmentofbiomedicalequipment
forpalliativecare. Section3willoutlineourmaindevelopmentsrelatedtoflowmonitoringbased
ontheDopplereffect. Wewilldiscussouradvancesinprocessingoftheflowsignals,thereal-time
monitoringofnon-steadyflows,bloodflowmeasurementsonhumanskinandOFI-basedpressure
myography. Thelastsectionwilldiscusstheconclusionsofthepaper.
2. OpticalPathChanges
2.1. OFIfortheMonitoringofTransientPathChanges
2.1.1. Introduction
Theapplicationoftransientopticalpathchanges(forcedorfreechangesineitherdistanceorthe
refractiveindexofthemediacrossedbythebeam,whosestatisticsvarywithtime)isusedextensively
in biosensing and medical signals, usually related to vibration (e.g., in electroencephalography
(EEG), electrocardiography (ECG) or vibrocardiography (VCG)) [36–39]. So far, optical feedback
interferometry(OFI)hasbeennormallyusedtoanalyzeperiodicvibrationpatternsusingdifferent
signalprocessingschemes(outlinedintheIntroduction)toextractvibrationalparameters,suchas
displacement,frequencyandtheaveragevelocityofvibratingtargets. WhileFourier-basedsignal
processing is an elegant approach, it requires the measurement of the complete signal over time
andcannotbeusedinordertoanalyzetransientsignalsorsignalswithunknownfuturebehavior.
Fringecounting,thoughbeingsimpleinordertoprocessOFIsignals,isnotsuitablewhensignals
aretime-dependent,asthethresholdofdetectionoftenneedstobechanged,optimizedoradapted
dependingonthedisplacementofthetarget.
Here,wepresentamethodologyfortheanalysisoftransientvibrationsusinganOFIsensorbased
onanadaptedprocessingstrategy,hencebroadeningthescopeofthemeasurementandmakingOFI
sensorsmoresuitablefortheiruseasbiosensor. Toharnessthepoweroffrequencyspectrumanalysis
without the need of a complete characterization of the signal in time, wavelet analysis provides
a very suitable framework [13,40]. In such a representation, the frequency domain reflects the
behaviorofatemporally-localizedversionofthesignal,thusmakingitsuitedtostudyatransientor
undeterminedsignal.
2.1.2. ExperimentalSetup
TheexperimentalsetupispresentedinFigure1. AHL8337MGAlGaAslaserdiode(LD)was
used. Theemissionwavelength,measuredwithanInstrumentSystem’sSPECTRO320(D)R5unit,
wasmeasuredasλ=692.5nm. TheopticalbeamemittedfromtheLDwasfocusedusingaThorlabs
lens 352,240 with a focal length of 8 mm and numerical aperture of 0.5 on target. The target was
apiezoelectric(PZT)linearstagePhisikInstrumente-LISA(PI-LISA)(P-753.3CD),whichincludedan
embeddedcapacitivesensorwitharesolutionof0.2nm. Thevoltagetoamplitudeconversionfactor
was3.8µm/V.ToanalyzethetransientresponseusingOFI,asignalintheformofanon-periodic
sincpulsewasappliedtothePZT(Figure2a). The3-dBpulseduration(timedurationforwhichthe
Sensors2016,16, 694 4of26
√
voltageofpulsebecomes1/ 2ofthemaximumvalue)wasτ = τ −τ =613.3−592.8=20.5ms,
3dB 1 2
whereτ andτ arethefirstandsecond3-dBpoints. Thepeakvalueofthepulseoccursatτ =603ms,
1 2 0
whereitpresentsamaximumamplitudeofV =1V.Thisperturbationinvoltageinducedatransient
0
vibrationofamplitudeof A = 3.8µm= 5λ/2onthePZT.Theperturbationisthenmeasuredby
m
theOFIsensordescribedandtheoscillationsignalfromtheinternalmonitorphotodiodemeasured
inanoscilloscopeconnectedtoacomputer,asinFigure2b. Itisobservedthat,asmostofthetime,
thevoltagesignalappliedtoPZTiskeptconstant,novibrationoccursinthePZT,andnofringesare
registered. Atthetimemomentt = τ = 603ms,thevoltageswitchestoV = 1V,andthefringes
0 0
areproducedandcaptured(Figure2a). ThenumberoffringesdetectedN =6isconsistentwiththe
f
amplitudeofvibration[9]. Theoscilloscopedigitizestheanalogsignalfromthephotodiodeandsends
thedigitizedsignaltoacomputer,wherefurtherprocessingisperformed.
PD LD Piezo
Oscilloscope
Figure1.Experimentalsetupforthemonitoringoftransientpathchanges.
(a)
(b)
Figure2. Experimentalopticalfeedbackinterferometry(OFI)signal. (a)Transientsignalappliedto
PZTtointroducetransientvibration;(b)OFIsignalresultingfromtransientmotionofthepiezo;the
insetgivesthemagnifiedviewofOFIsignalproducedbytransientvibration.
2.1.3. Results
Havingshownthatthetransientvibrationofthetarget(PZT)iseffectivelydetectedusingtheOFI
signal,theuseofwaveletsandalgorithmstoprocesstheOFIsignalmaybedirectlyapplied(see[13,40]
Sensors2016,16, 694 5of26
foradetaileddescription).WeproposeanalgorithmbasedontheMorletwavelet,whichenablesoneto
simultaneouslyobtainspectralandtemporalinformationrelatedtothetransientvibration,including
itsdurationandtheinstantaneousvelocityprofileintroduced. Wavelettransformsarefunctionsthat
are finite both in space and time, enabling one to retrieve information both on the temporal and
frequencyaspectsofthesignalthroughthewaveletcoefficientsW(a,b). Foracontinuoussignalx(t),
W(a,b)aregivenby:
1 (cid:90) +∞ t−b
W(a,b) = √ x(t)ψ( )dt (1)
a −∞ a
withW(a,b)thecoefficientofwavelettransformationandaandbthedilation(scaling)andtranslation
(shifting)parameters. TheMorletwaveletisdefinedinthetimedomainas:
−t2
ψ(t) =exp(jω t)exp( ) (2)
o
2
wherej2 = −1,andω issetto5.336[41].
o
Inourapproach,acontinuouswavelettransform(CWT)hasbeencalculateddirectlyonthesignal,
allowingbothnoiseremovalanddetectionoftherelevantparametersofthesignalinasinglestep.
SuchcalculationsareperformedusingaMATLABcode,typicallyinafewseconds,soitsreal-time
implementation is feasible. Figure 3a shows the captured OFI signal, while Figure 3b shows the
associatedscalogramoftheOFIsignal. Thescalogramrepresentsthecalculatedwaveletcoefficients
inthetime-frequencyplaneW(f,t),where f andtstandforfrequencyandtime,respectively. Itis
observed that for most of the time of the experiment, the wavelet coefficients stay close to zero
(representingtheabsenceofaspectralcomponent). Duringthepulseperiod,however,theirvalues
changesignificantly,demonstratingtheriseofnovelspectralcomponentsresultingfromtheDoppler
shiftalongthepulseduration. FromFigure3b,theDopplerfrequencyis f =453.72Hz.
d
(a)
(b)
Figure 3. Time-Frequency representation of the OFI signal. (a) OFI signal; (b) scalogram of the
OFIsignal.
Sensors2016,16, 694 6of26
Tofindthe3-dBpulsedurationoftransientvibration,thewaveletcoefficientsat f = f =453.72Hz
d
foralltime,i.e.,W(f )areextractedfromthescalogram. Forsimplicity,onlythewaveletcoefficientsin
d
thevicinityofthepulsearepresentedinFigure4. ItisobservedthatthevalueofW(f )issignificant
d
onlyduringthepulsedurationandiszeroanywhereelse. TheplotofW(f )hastwomaximaatτ(cid:48) and
d 1
τ(cid:48) withasharpdipatτ(cid:48) representedbyPointsA,BandC,respectively,inFigure4. The3-dBwidth
2 0
(τ(cid:48) )andcenterofpulse(τ(cid:48))arecalculatedfromFigure4as:
3dB 0
τ(cid:48) = τ(cid:48)−τ(cid:48) =613.3−592.5=20.8ms (3)
3dB 2 1
τ(cid:48) =603ms (4)
0
showingverygoodcorrespondencebetweentheinducedperturbationandthemeasurementextracted
fromtheOFIsignal,asmaybeobservedinTable1.
1 A (τ’ ) C (τ’2)
1
0.9
W(f )
d
0.8
transient vibration
0.7
u) 0.6
a.
e (
d 0.5
u
plit
m 0.4
A
0.3 τ’ =592.5 msec
1
τ’ =613.3 msec
0.2 2
τ’ =603 msec
0
τ’ =τ’ −τ’ =20.8 msec
0.1 3dB 2 1
B (τ’ )
0
540 560 580 600 620 640 660
Time (msec)
Figure4.Experimentalresults.Characterizingthepulse:determiningthe3-dBwidthandcentertime
ofthepulse.
Table1.Comparisonbetweenthereferenceandmeasuredpulseparameters.
Parameters OriginalValue CalculatedValue Error
Centerpulsetime τ =603ms τ(cid:48) =603ms 0.0%
0 0
First3-dBtime τ =592.8ms τ(cid:48) =592.5ms 0.05%
1 1
Second3-dBtime τ =613.3ms τ(cid:48) =613.3ms 0.0%
2
3-dBpulseduration τ =τ −τ =20.5ms τ(cid:48) =τ(cid:48)−τ(cid:48) =20.8ms 1.4%
3dB 2 1 3dB 2 1
Peakvelocity V =1.65×10−4mm/ms V(cid:48) =1.57×10−4mm/ms 4.8%
0 0
Finally, we will determine the velocity profile v(t) of the displacement from the scalogram.
First,the maximum wavelet coefficient present at each instant of time is obtained from Figure 3.
Then,thecorrespondingfrequencyisfound, whichgivesthevalueoftheDopplerfrequencyshift
asafunctionoftime f (t). Oncetheinstantaneousfrequencyisknown,thevelocityofthetargetis
d
obtained using v(t) = f (t)λ/2 and reconstructed in time in Figure 5. The amplitude plot of the
d
signalusedfortheexcitationofthepiezohasbeenplottedtoenablecomparison. Itisobservedthat
thecalculatedpeakvelocityofthetargetisV(cid:48) = 1.57×10−4 mm/ms,whichmaybecomparedto
0
Sensors2016,16, 694 7of26
thatoftheactual,ofV = 1.65×10−4 mm/ms(fromFigure5,showingalsothatthetargetmakes
0
a displacement of 3.8 µm in 23 ms). In addition, a sharp dip in the velocity profile is observed
consequenttothefactthatthetargethasreachedthemaximumvalueoftheperturbationandnow
vibratesintheoppositedirection. Asthesignalappliedtothepiezocorrespondstoasincfunction
withtime,somebiashasbeenaddedtotheamplitudesignaltoavoidnegativepolarizationofthePZT.
Theoscillationsaroundthepeakareduetotheshapeofthesincfunctionandarerecoveredinthe
velocityprofile. Thecombinationofthebiasandtheoscillationsofthesincfunctionresultinasmall
velocitycomponentduetotheDopplershift.
x 10−4
4
transient vibration
v(t)
1.75 3.5
V’ =1.57 × 10−4 mm/msec
0
1.5 3
c)
se 1.25 2.5m)
m µ
m/ e (
m 1 2 d
city ( mplitu
elo 0.75 1.5A
V
0.5 1
0.25 0.5
0 0
440000 445500 550000 555500 660000 665500 770000 775500 880000 885500
Time (msec)
Figure5.Experimentalresults.Velocityprofileofthetransientpulsedeterminedusingwavelets(blue)
andtime-dependentsignalappliedtothetarget(red).Circlesindicatetheaxisforeachfigure.
The method performs very accurately regarding the detection of time events, providing
aconvenientmethodfortheanalysisoftime-dependentphenomenainOFIsignals. Thus,complete
and accurate characterization of transient vibrations using a Morlet wavelet approach has been
demonstrated, presentinganovel, usefultoolforthedetectionandanalysisofbiomedicalsignals,
includingthecharacterizationofitstimeandspeedsignatures.
2.2. OFISensorforLargeDisplacements
Self-mixinghasbeenusedfordisplacementsensing[42],thoughthepresenceofspecklenoise,
whichcausesvariationsinboththeamplitudeofOFIsignalsandthevalueoftheopticalcoupling
factorC, has often reduced the range or even the possibility of correct measurement. Different
approacheshavebeendevelopedtocopewithspeckleeffects,suchas:
• Usingtwopiezo-actuatorstomovealensforaspeckle-trackingtechnique[43],
• Usingavoltage-controlledliquidlensandadouble-headedlaserdiodesensorwithdifferentlaser
beamspotsizeshavealsobeenproposedtoavoidspeckle[44],
• KeepingtheoperatingpointoftheOFIinterferometerfixedatthehalf-fringe[45],
• UsingOFIsignalenvelopetrackinganddynamicfringedetection[46],
• UsingtheHilberttransformand/orthewavelettransformtodetect[12,13].
Only the last three methods demonstrate signal processing techniques that correctly extract
andprocessOFIsignalscorruptedbyspeckle,whileavoidingadditionaloptical/electro-mechanical
Sensors2016,16, 694 8of26
components. However,themethodbasedonfeedbacktechniques[45]hasalimitedrangeofoperation
(<100µm)comparedtothelasttwoopenloopapproaches.
Here, as an example, the OFI technique has been successfully applied to quantify the
displacementsoftheGoogleLiftwarespoon(Figure6),inordertodetermineitsefficiencytodamp
hand-shaking vibration due to diseases, such as Parkinson. The experimental setup is shown in
Figure6. Forthesamegivenfrequencyandamplitudevibration,boththedisplacementofthehandle
andofthestabilizedutensilaremeasuredtoestimatetheLiftwaresystemefficiency. Notealsothatthe
Liftwareisplacedontopofasteelpostinordertoreducetheparasiticinfluenceofthemagnetofthe
shaker. Duetothenon-linearityoftheinducedmovementandtheamplitudeofthedisplacements
(a few millimeters), the OFI signals obtained with an oscilloscope from the internal photodiode
areunavoidablyaffectedbyadynamically-changingcouplingfactor, plusthepresenceofspeckle,
asshowninFigure7. Consequently,inordertodetectthesecorruptedOFIfringes,themethodused
hereisbasedontheHilberttransform. TheresultsaresummarizedinFigure8. Thebiggestinduced
displacement(4.56mm)onthehandlewasobtainedat4Hz,whilethesmallest(2.48mm)at15Hz.
Dependingontheshakingfrequency,itcanbeobservedthatthestabilizinghandlecaneffectively
suppressvibrationsfrom25%(at4Hz)upto80%(at15Hz)oftheutensil. Whilefurtherworkis
necessarytoproperlycharacterizetheLiftware’stransferfunction,notably,intermsoftheinduced
movement,itcanbeforeseenthattheOFIconfigurationcoupledtoarobustsignalprocessingisan
adaptedlow-costsolutionforthistask.
OFI
displacement
sensor
Liftware
Liftware utensil stabilizing handle
Shaker
Figure6.Experimentalsetuptomeasurethevibrationofthestabilizinghandleandthevibrationof
theutensilattachment.
V)
e ( 0.05
d
u
plit 0
m
a
al −0.05 c)
n
g
si
FI −0.1 b)
O
0.12 0.14 0.16 0.18 0.2 0.22 0.24 0.26 0.28
Time (s)
(a)
Figure7.Cont.
Sensors2016,16, 694 9of26
0.04
V) 0.1 V)
de ( 0.05 de ( 0.02
u u
plit 0 plit
m m 0
al a−0.05 al a
sign −0.1 sign−0.02
FI FI
O−0.15 O−0.04
0.1635 0.164 0.1645 0.165 0.251 0.252 0.253 0.254 0.255
Time (s) Time (s)
(b) (c)
Figure7.MeasuredOFIsignalobtainedusingtheexperimentalsetupdescribedinFigure6exhibiting
strongvariationsinbothamplitudeanopticalcouplingfactorC.(a)Fullacquisitionthathighlightsthe
strongamplitudemodulationinducedbythespecklephenomenon;(b)truncationofthetime-domain
signalshowingahighCvalue;(c)truncationofthetime-domainsignalshowingalowervalueforthe
opticalcouplingfactorC.
Figure8.Measuredvibrationdisplacementofthestabilizinghandleandthevibrationoftheutensil
attachmentinducedbyashakeratdifferentstimulifrequencies.
3. OFIfortheMonitoringofFlows
3.1. ProcessingofOFIFlowSignalsunderDifferentScatteringRegimes
3.1.1. IntroductiontoSignalProcessingforFlowMonitoring
InOFI-basedflowmetry,theDopplerspectrumisanalyzedtoobtaintheinformationregardingthe
velocityofmovingparticles. SincetheDopplerspectrumisgeneratedfromthefeedbackofparticles
embedded in the fluid, its morphology strongly depends on the velocity distribution of particles
inside the sensing volume. Thus, the number of scattering particles in the sensing volume is also
animportantfactorinspectrumdistribution. Dependingontheparticleconcentrationinthefluid,
singlescatteringormultiplescatteringmayoccurinthesensingpoint,causingasignificantchangein
signalpowerspectrumoftheOFIsensor. Thereby,basedonthetypeofpowerspectrummorphology
(i.e.,narrowpeak,flatdistributionoraslowdecay),differentsignalprocessingmethodshavebeen
recentlyproposedtoaccuratelyextracttheDopplerfrequencyfromthepowerspectrumcorresponding
to the average fluid velocity at the measurement volume. The commonly-used methods include