Professional Documents
Culture Documents
ho
ng
BiomedicalSignalProcessing
r.R
.B
.G
Lecture0
INTRODUCTION
Dr.R.B.Ghongade
DepartmentofE&TC
Vishwakarma InstituteofInformationTechnology,Pune
INDIA
Syllabus
r.R
.B
.G
ho
ng
ad
e
ad
e
Texts/References
r.R
.B
.G
ho
ng
3.
5.
ng
ho
4.
.G
2.
Jaakko Malmivuo;RobertPlonsey;Bioelectromagnetism :
PrinciplesandApplicationsofBioelectricandBiomagnetic
Fields,OxfordUniversityPress
Antoun Khawaja
;AutomaticECGAnalysisusingPrincipal
ComponentAnalysisandWaveletTransformation,Karlsruhe
TransactionsonBiomedicalEngineering,2006
JohnL.Semmlow;Biosignal andBiomedicalImageProcessing:
MATLABBasedApplications,MarcelDekker,Inc.,2004
Rezaul Begg;Joarder Kamruzzaman;Ruhul Sarker;Neural
networksinhealthcare:potentialandchallenges,IdeaGroup
Publishing,2006
Lief Sornmo;PabloLaguna;BioelectricalSignalProcessingin
CardiacandNeurologicalApplications,Elsevier,2005,First
Edition
r.R
.B
1.
ad
e
AdditionalTexts/References
LecturePlan
r.R
.B
.G
ho
ng
ad
e
Lecture1:IntroductiontoBiomedicalInstrumentationand safety
considerations
Lecture2:Biopotentials
Lecture3:BioelectrodesandPhysicalMeasurements
Lecture4a:CardiovascularSystem
Lecture4b:Phonocardiography,EEG
Lecture5:XRayImaging,ComputedTomography,DiagnosticUltrasound
Imaging
Lecture6:AnalogSignalProcessingofBiosignals(NOPPT)
Lecture7:DigitalsignalprocessingofBiosignals
Lecture8:Software basedmedicalsignaldetectionandpatternrecognition
CaseStudy(NOPPT)
Assignments
ho
ng
ad
e
PARTI
Assig.1:SurveyofBiomedicalsensors
Assig.2:Designandsimulationofinstrumentationamplifier
Assig.3:DesignandsimulationofActiveFilters
r.R
.B
.G
PARTII
Assig.4:MATLABexercise(basicoperations,commands)
Assig.5:ECGSignalprocessingusingFFTandWavelets
Assig.6:ECGPatternclassification
Lecture1
r.R
.B
.G
ho
ng
ad
e
PARTI:IntroductiontoBiomedical
Instrumentation
PARTII:Safetyconsiderations
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
BiomedicalInstrumentation
ad
e
Amedicaldeviceis
.G
ho
ng
anyitempromotedforamedicalpurposethat
doesnotrelyonchemicalactiontoachieveits
intendedeffect
r.R
.B
Differencefromanyconventionalinstrument
sourceofsignalsislivingtissue
energyisappliedtothelivingtissue
Howdoesthisimpactdesignrequirements?
Reliability,Reliability,Reliability!!!
r.R
.B
.G
ho
ng
ad
e
Timelineofmajor
inventions
r.R
.B
.G
ho
ng
ad
e
Generalizedinstrumentationsystem
Measurand
r.R
.B
.G
ho
ng
ad
e
Physicalquantity,propertyorconditionthatthe
systemmeasures
Typesofmeasurands
InternalBloodpressure
BodysurfaceECGorEEGpotentials
PeripheralInfraredradiation
OfflineExtracttissuesample,bloodorbiopsy
Categoriesofmeasurands
Biopotential,pressure,flow,dimensions
(imaging),displacement(velocity,acceleration
andforce),impedance,temperatureand
chemicalconcentration
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
Sensor
r.R
.B
.G
ho
ng
ad
e
Asensorconvertsphysicalmeasurand toan
electricaloutput
Sensorrequirements
Selective shouldrespondtoaspecificform
ofenergyinthemeasurand
Minimallyinvasive shouldnotaffectthe
responseofthelivingtissue
Mostimportanttypesofsensorsinbiomedical
systems
displacement
pressure
SignalConditioning
r.R
.B
.G
ho
ng
ad
e
SignalConditioning:Amplificationandfiltering
ofthesignalacquiredfromthesensortomake
itsuitableforprocessing/display
Generalcategories
Analog,digitalormixedsignal
Timedomainprocessing
Frequencydomainprocessing
Spatialdomainprocessing
OperationalModes
Directvs.Indirect
r.R
.B
.G
ho
ng
ad
e
Directmode:measuredesiredmeasurand directly
ifthesensorisinvasive,directcontactwiththe
measurand ispossiblebutexpensive,riskyandleast
acceptable
Indirectmode:measureaquantitythatisaccessibleand
relatedtothedesiredmeasurand
assumption:therelationshipbetweenthemeasurands is
alreadyknown
oftenchosenwhenthemeasurand requiresinvasive
procedurestomeasuredirectly
Exampleindirectmode
Cardiacoutput(volumeofbloodpumpedperminutebythe
heart)canbedeterminedfrommeasurementofrespiration,
bloodgasconcentration&dyedilution
Organmorphologycanbedeterminedfromxrayshadows
OperationalModes
Samplingvs.Continuousmode
ho
ng
ad
e
Sampling:forslowvaryingmeasurands thataresensed
infrequentlylikebodytemperature&ionconcentrations
Continuous:forcriticalmeasurementsrequiringconstant
monitoringlikeelectrocardiogramandrespiratorygas
flow
.G
Generatingvs.Modulating
r.R
.B
Generating:alsoknownasselfpoweredmodederive
theiroperationalenergyfromthemeasurand itself
Example:piezoelectricsensors,solarcells
Modulating:measurand modulatestheelectricalsignal
whichissuppliedexternallymodulationaffectsoutputof
thesensor
Example:photoconductiveorpiezoresistive sensor
MeasurementConstraints
r.R
.B
.G
ho
ng
ad
e
Otherthandevicefunctionality,thesignaltobemeasured
imposesconstraintsonhowitshouldbeacquiredand
processed
Measurementandfrequencyranges
Mostmedicalmeasurands aretypicallymuchlowerthan
conventionalsensingparameters(microvolts,mmHg,low
frequency)
Interferenceandcrosstalk
Notpossibletoisolateeffectsofothermeasurands
CannotmeasureEEGwithoutinterferencefromEMG
Placementofsensorsandcompensation/calibration
processplayakeyroleinanybioinstrumentationdesign
MeasurementConstraints
Measurementvariabilityisinherentatmolecular,organandbodylevel
Primarycause
interactionbetweendifferentphysiologicalsystems
existenceofnumerousfeedbackloopswhosepropertiesare
poorlyunderstood
Thereforeevaluationofbiomedicaldevicesrelyonprobabilistic/statistical
methods(biostatistics)
SAFETY
Duetointeractionofsensorwithlivingtissue,safetyisaprimary
considerationinallphasesofthedesign&testingprocessthe
damagecausedcouldbeirreversible
Inmanycases,safelevelsofenergyisdifficulttoestablish
Safetyofmedicalpersonnelalsomustbeconsidered
Operatorconstraints
Reliable,easytooperate,ruggedanddurable
r.R
.B
.G
ho
ng
ad
e
Classificationofbiomedicalinstruments
r.R
.B
.G
ho
ng
ad
e
Quantitybeingsensed
pressure,flowortemperature
makescomparisonofdifferenttechnologieseasy
Principleoftransduction
resistive,inductive,capacitive,ultrasonicor
electrochemical
makesdevelopmentofnewapplicationseasy
Organsystems
cardiovascular,pulmonary,nervous,endocrine
isolatesallimportantmeasurementsforspecialists
whoneedtoknowaboutaspecificarea
Clinicalspecialties
pediatrics,obstetrics,cardiologyorradiology
easyformedicalpersonnelinterestedinspecialized
equipment.
MeasurementInputSources
r.R
.B
.G
ho
ng
ad
e
Desiredinputs
measurands thattheinstrumentis
designedtoisolate
Interferinginputs
quantitiesthatinadvertentlyaffectthe
instrumentasaconsequenceofthe
principlesusedtoacquireandprocess
thedesiredinputs
Modifyinginputs
undesiredquantitiesthatindirectly
affecttheoutputbyalteringthe
performanceoftheinstrumentitself
ECGexample
Desiredinput ECGvoltage
Interferinginput 50Hznoisevoltage,
displacementcurrents
Modifyinginput orientationofthepatient Interferinginputsgenerallynot
cables
correlatedtomeasurand
whentheplaneofthecableis
ofteneasytoremove/cancel
perpendiculartothemagnetic
Modifyinginputsmaybecorrelatedto
fieldthemagneticinterferenceis
themeasurand
maximal
moredifficulttoremove
r.R
.B
.G
ho
ng
ad
e
DesignCriteriaandProcess
RegulationofMedicalDevices
r.R
.B
.G
ho
ng
ad
e
Regulatorydivisionofmedicaldevices:classI,IIandIII
moreregulationfordevicesthatposegreaterrisk
ClassI(Generalcontrols)
Manufacturersarerequiredtoperformregistration,
premarketingnotification,recordkeeping,labeling,
reportingofadverseexperiencesandgood
manufacturingpractices
ClassII(Performancestandards)
800standardsneededtobemet!
ClassIII(Premarketingapproval)
Manufacturershavetoprovethesafetyofthesedevices
priortomarketrelease
Implanteddevices(pacemakersetc.)aretypically
designatedclassIII
Investigationaldevicesaretypicallyexempt
CompensationTechniques
r.R
.B
.G
ho
ng
ad
e
Compensation:eliminationorreductionofinterferingand
modifyinginputs
Techniques
Alteringthedesignofessentialinstrumentcomponents
simpletoimplement
Addingnewcomponentstooffsettheundesiredinputs
Methods
Reducesensitivitytointerferingandmodifyinginputs
Example:usetwistedcablesandreducenumberof
electricalloops
SignalFiltering
temporal,frequencyandspatialseparationofsignal
fromnoise
Compensation:NegativeFeedback
r.R
.B
.G
ho
ng
ad
e
Whenmodifyinginputcannotbeavoided,negativefeedbackisused
tomaketheoutputlessdependentonthetransferfunctionofthe
device
Feedbackdevicesmustbeaccurateandlinear
ad
e
ng
ho
.G
r.R
.B
Feedback
ad
e
OtherCompensationTechniques
Opposinginputsornoisecancellation
ho
ng
Wheninterferingandmodifyinginputs
cannotbefiltered
r.R
.B
.G
additionalinputscanbeusedtocancel
undesiredoutputcomponents
similartodifferentialsignalrepresentation
Biostatistics
ad
e
Usedtodesignexperimentsandclinical
studies:
r.R
.B
.G
ho
ng
Tosummarize,explore,analyzeandpresentdata
Todrawinferencesfromdatabyestimationorby
hypothesistesting
Toevaluatediagnosticprocedures
Toassistclinicaldecisionmaking
Medicalresearchstudiescanbeclassifiedas:
Observationalstudies:Characteristicsofoneormore
groupsofpatientsareobservedandrecorded.
Experimentalinterventionstudies:Effectofamedical
procedureortreatmentisinvestigated.
BiostatisticsStudies
Observationalstudies caseseriesstudies
.G
Crosssectionalstudies:
ho
ng
ad
e
Casecontrolstudies
useofindividualsselectedbecausetheyhavesome
outcomeordisease
thenlookbackwardtodeterminepossiblecauses
r.R
.B
Analyzecharacteristicsofpatientsatoneparticulartime
todeterminethestatusofadisease orcondition.
Cohortobservationalstudies:
Aparticularcharacteristicsisaprecursorforanoutcome
ordisease
Controlledstudies:
Ifprocedurescomparedtotheoutcomeforpatients
givenaplaceboorotheracceptedtreatment
ad
e
BiostatisticsStudiesII
ng
Concurrentcontrol:
r.R
.B
Doubleblindstudy:
.G
ho
Patientsareselectedinthesamewayandforthesame
duration
Randomizedselectionofpatientstotreatmentoptions
tominimizeinvestigatororpatientbias
Biostatistics:DataAnalysis
.G
ho
ng
ad
e
Distributionsofdatareflectthevaluesofavariable/
characteristicandfrequencyofoccurrenceofthosevalues
r.R
.B
Mean:(X)averageofNvalues(arithmeticorgeometric
mean)
Median:middleofrankedvalues
Mode:mostfrequentvalue
ad
e
Biostatistics:DataAnalysis
ho
ng
Standarddeviation:(s)spreadofdata
75%ofvaluesliebetween
r.R
.B
.G
CoefficientofVariation:(CV)
permitscomparisonofdifferentscales
Percentile
Percentageofdistributionthatislessthanor
equaltothepercentilenumber
MoreBiostatistics
ng
ho
ad
e
Correlation coefficient(r)
.G
EstimationandHypothesisTesting
r.R
.B
Confidenceintervals
indicatesthedegreeofconfidencethatdatacontainsthetruemean
Hypothesistesting
revealswhetherthesamplegivesenoughevidenceforustorejectthe
nullhypothesis(statementexpressingtheoppositeofwhatwethinkis
true)
Pvalue:
howoftentheobserveddifferencewouldoccurbychancealone
MoreBiostatistics
ad
e
Methodsformeasuringtheaccuracyofadiagnostic
procedure:
r.R
.B
.G
ho
ng
Sensitivity:probabilityofthetestyieldingpositiveresultsinpatients
whoactuallyhavethedisease
opposite:falsenegativerate
Specificity:probabilityofthetestyieldingnegativeresultsinpatients
whodonothavethedisease
opposite:falsepositiverate
InstrumentCharacterization
ng
ad
e
Enablecomparisonofavailableinstruments
Permitevaluationofnewinstrumentdesigns
ho
Generalizedstaticcharacteristics
r.R
.B
.G
Staticcharacteristics:
performanceofinstrumentsfordcorverylow
frequencyinputs
somesensorsrespondonlytotimevaryinginputsand
havenostaticcharacteristics
Dynamiccharacteristics:
requiretemporalrelationshipstodescribethequality
ofmeasurements
StaticCharacteristics
ad
e
Accuracy
.G
ho
ng
Differencebetweenthetruevalueandthemeasuredvalue
normalizedbythemagnitudeofthetruevalue
Severalwaystoexpressaccuracy
mostpopularisintermsofpercentageoffullscale
measurement
r.R
.B
Precision
Resolution
Expressesnumberofdistinguishablealternativesfromwhichagiven
resultisselected
Highprecisiondoesnotmeanhighaccuracy.
Smallestincrementalquantitythatcanbemeasuredwithcertainty
Reproducibility
Abilityofaninstrumenttogivethesameoutputforequalinputs
appliedoversomeperiodoftime
StatisticalControlandStaticSensitivity
ad
e
Measurementconditionshavetotakeintoaccountrandomnessintroducedby
environmentalconditions
Ifthesourceofvariationcannotberemoved,thenuseaveraging
.G
ho
ng
Statisticsensitivity(dcgain)
Toperformcalibrationbetweenoutputandinput
Forlinearcalibration
r.R
.B
Astaticcalibrationisperformedbyholdingallinputs(desired,interfering,
andmodifying)constantexceptone
Thisoneinputisvariedincrementallyover
thenormaloperatingrange,resultingina
rangeofincrementaloutputs.
Thestaticsensitivityofaninstrumentor
systemistheratiooftheincremental
outputquantitytotheincrementalinput
quantity
StaticCharacteristics
Zerodrift(offseterror)
Whenallmeasurements
increasesordecreasebythe
sameabsoluteamount
Causes:manufacturing
misalignment,variationsin
ambienttemperature,
hysteresisvibration,shock,dc
offsetvoltageatelectrodes
Sensitivitydrift(gainerror)
Whentheslopeofthe
calibrationcurvechangesasa
resultofinterferingor
modifyinginput
Causes:manufacturing
tolerances,variationsinpower
supply,nonlinearity
Example:ECGamplifiergainchanges
duetodcpowersupplyvariation
r.R
.B
.G
ho
ng
ad
e
Linearity
r.R
.B
.G
ho
ng
ad
e
Linearity:Asystemthatdemonstratessuperpositionprinciple
Measureoflinearity:
a)maximaldeviationofpointsfromtheregressionline
expressedaspercentageofthefullrangeor
b)harmonicdistortionmeasure.
MoreStaticCharacteristics
Inputranges
.G
Inputimpedance(Z)
ho
ng
ad
e
Constraintsonlinearityimposesanoperationalrangefortheinput
parameters
Inputrangeisalsoapplicabletointerferinginputs(usedforshieldingof
instruments)
r.R
.B
Measuresthedegreetowhichinstruments
disturbthequantitybeingmeasured
effortvariable:examplesvoltage,pressure,
force
flowvariable:examplescurrent,flow,velocity
whenmeasuringeffortvariables,inputimpedance
shouldverylarge
whenmeasuringflowvariables,inputimpedance
shouldverysmall
DynamicCharacteristics
r.R
.B
.G
ho
ng
ad
e
Quantifyresponseofmedicalequipmentwithrespectto
timevaryinginputs
Manyengineeringinstrumentscanbedescribedbyordinary
lineardifferentialequations
Mostpracticalinstrumentshaveafirstorsecondorder
response
Practicalevaluationofasystem
Applyinputasaunitstepfunction,sinusoidalfunctionor
whitenoise
DynamicCharacteristics
ho
ng
ad
e
Operationaltransferfunction:
r.R
.B
.G
Frequencyresponseofasystem
Forasinusoidalinput
theoutputisasinusoidwithdifferentmagnitudeand
phase
Magnitude:
Phase:
.G
r.R
.B
Linear
potentiometerisan
exampleofazero
orderinstrument
Inpractice,athigh
frequencies
parasitic
capacitanceand
inductancewill
causedistortion
ho
ng
ad
e
ZeroorderInstrument
FirstorderInstrument
r.R
.B
.G
Staticsensitivity(dcgain):
ho
ng
ad
e
Firstorderinstrumentcontainsasingleenergystorage
element
Timeconstantofthesystem:
Afrequencytransferfunctionisgivenby
r.R
.B
.G
ho
ng
ad
e
FirstorderInstrument
SecondorderInstrument
.G
ho
ng
ad
e
Secondorderinstrumentcontainsaminimumoftwo
energystorageelement
r.R
.B
Staticsensitivity(dcgain):
Undampednaturalfrequency:
Dampingratio:
r.R
.B
.G
ho
ng
ad
e
SecondorderInstrument
ad
e
ng
ho
r.R
.B
.G
MedicalInstrumentElectricalSafety
Significanceofsafety
r.R
.B
.G
ho
ng
ad
e
Tensofthousandsdevicerelatedpatient
injuriesinU.Severyyear.
Evenasingleharmfuleventcanleadto
significantdamageintermsofreputationand
legalaction.
Differentlevelofprotectionrequiredas
comparedtohouseholdequipment.
Minimumperformancestandardsintroduced
in1980srelativelynewpractice.
r.R
.B
.G
ho
ng
ad
e
PhysiologicalEffectsofElectricity
Experimentsfrom160lbhumanwith60Hzcurrent
ho
1.1mAformen
0.7mAforwomen
ng
Meanthresholdofperception
ad
e
SusceptibilityParameters
.G
Minimumthresholdofperception500A
r.R
.B
80A withgelelectrodes(reducesskinimpedance)
Meanletgocurrent
16.5mAformen
10.5mAforwomen
Letgocurrentvs.frequency
Minimalletgocurrentoccursatcommercialpowerlinefrequenciesof
5060Hz
ad
e
ng
ho
.G
r.R
.B
D
FromexperimentsperformedbyCharlesDalziel (1940to1950
Shock(stimulation)duration
ad
e
SusceptibilityFactors
ng
Fibrillationcurrentisinverselyproportionaltotheshockpulseduration
longerpulses>lowercurrentdoesdamage
ho
Bodyweight
.G
Fibrillationcurrentincreaseswithbodyweight
r.R
.B
50mARMSfor6Kgdogs
130mARMSfor24Kgdogs
Pointsofentry
Skinimpedancevaries:15k to1M
Resistivebarrierthatlimitscurrentflow
Tissue(beneathskin)haslowimpedance
Macrovs.MicroShock
Macroshock
ad
e
externallyappliedcurrent
spreadsthroughthebodysolessconcentrated
ng
Microshock
r.R
.B
.G
ho
appliedcurrentisconcentratedataninvasivepoint
acceptedsafetylimitisonly10A
generallyonlydangerousifcurrentflowsthroughtheheart
Macroshock Hazards
Mostprobablecauseofdeathduetomacroshock
ventricularfibrillation
Factors
skin/bodyresistance
designofelectricalequipment
Skinandbodyresistance
dryskinhashighresistance(~15k1Mohm)
limitscurrentthroughbody
wet/brokenskinhaslowresistance(~1%thatofdryskin)
internalbodyresistance
~200ohmforeachlimb
~100ohmfortrunkofbody
resistancebetweentwolimbs=~500ohm
proceduresthatbypassskinresistancecanbedangerous
example:gelelectrodes,surgery,oral/rectalthermometers
r.R
.B
.G
ho
ng
ad
e
Microshock Hazards
Maincauses
ho
ng
ad
e
leakagecurrentsinlineoperatedequipment
undesiredcurrentsthroughinsolatedconductorsatdifferent
potentials
differencesinvoltagebetweengroundedconductivesurfaces
.G
Leakagecurrents
r.R
.B
iflowresistancegroundisavailable>noproblem
ifgroundisbroken>currentflowsthroughpatient
ConductivePaths
ng
ad
e
Directconnectiontoaninternalorgan(during
measurementorsurgery)makespatientssusceptible
tomircoshock
r.R
.B
.G
ho
Externalelectrodesoftemporarycardiacpacemakers
Electrodesforintracardiacmeasuringdevices
Liquidfilledcathetersplacedintheheart
liquidfilledcathetershavemuchgreaterresistancethan
electrodes
Worst!danger!
currentsflowingthroughtheheart
Electrodecurrentdensity
experimentssuggestsmallerelectrodearemoredangerous
PowerDistribution
ElectricalpowersysteminHealthcareFacility
mustcontrolavailablepower(fuse/breakertosetmaxcurrent)
mustprovidegoodground
PatientsElectricalEnvironmentGrounding
NECcode:maxpotentialbetweentwosurfaces
generalcareareas:500mVundernormaloperation
criticalcareareas:40mVundernormaloperation
IsolatedPowerSystems
Groundfault
shortcircuitbetweenhotconductorandground
injectslargecurrentintogroundingsystem
cancreatehazardouspotentialsongroundedsurfaces
Isolationtransformer
isolatesconductorsagainstgroundfaults
mayincludegroundfaultmonitor/alarm
r.R
.B
.G
ho
ng
ad
e
GroundLoops
ng
ho
r.R
.B
.G
allintensivecareunitsmust
havesinglegroundforeach
patientisolatedfrom
hospitalground
40mVlimitonpotentialof
anyconductivesurfaces
ad
e
Differencesinground
potential:majorsourceof
microshock
Example:currentdueto
groundloopflowsthrough
patient
ElectricalIsolation
.G
Barrierisolation
r.R
.B
transformer,opticalorcapacitive
isolation
nocurrentacrossbarrier
Implants
ng
ho
devicesthatbreakohmiccontinuityof
electricsignalsbetweeninputand
outputoftheamplifier
differentsupplyvoltagesourcesand
differentgroundsoneachsideofthe
barrier
ad
e
Isolationamplifiers
properinsulationrequiredtoprevent
Reviewquestions
r.R
.B
.G
ho
ng
ad
e
1. Drawtheblockdiagramofatypicalbiomedical
instrumentationsystem.
2. Enlistthedesigncriteriaforbiomedicalinstrumentation
system.
3. Classifybiomedicalinstruments.
4. Enlistvariousphysiologicalprocesses/parametersandtheir
ranges.
5. Whatdoyoumeanbybiostatistics?
6. Whatarethecharacteristicsofabiomedicalinstrumentation
system?
7. Commentonthesafetyaspectsofabiomedicalsystem.
ad
e
ng
ho
.G
NextClassBiopotentials
r.R
.B
(Ithasalotofpotential!)
ad
e
Lecture2
r.R
.B
.G
ho
ng
Biopotentials
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
Cells
r.R
.B
.G
ho
ng
ad
e
Cellthebasicunitoflivingtissue
Specializedintheiranatomyandphysiologytoperformdifferent
tasks
Allcellsexhibitavoltagedifferenceacrossthecellmembrane.
Nervecellsandmusclecellsareexcitable
Theircellmembranecanproduceelectrochemicalimpulsesand
conductthemalongthemembrane.
Inmusclecells,thiselectricphenomenonisalsoassociatedwiththe
contractionofthecell
Inothercells,suchasglandcells,itisbelievedthatthemembrane
voltageisimportanttotheexecutionofcellfunction
Theoriginofthemembranevoltageisthesameinnervecellsasin
musclecells.
Inbothcelltypes,themembranegeneratesanimpulseasa
consequenceofexcitation.
Thisimpulsepropagatesinbothcelltypesinthesamemanner
ng
ho
.G
r.R
.B
PartsoftheNerveCell
Thenervecellmaybe
dividedonthebasisofits
structureandfunctioninto
threemainparts:
(1)thecellbody,alsocalledthe
soma
(2)numerousshortprocesses
ofthesoma,calledthe
dendrites
(3)thesinglelongnervefiber,
theaxon
ad
e
StructureofNerveCell
StructureofNerveCell
ad
e
ng
ho
.G
r.R
.B
TheCellMembrane
ad
e
ng
ho
.G
r.R
.B
Thecellisenclosedbyacellmembranewhosethicknessisabout7.5 10.0nm
Itsstructureandcompositionresembleasoapbubblefilm,sinceoneofitsmajor
constituents,fattyacids,hasthatappearance
Thefattyacidsthatconstitutemostofthecellmembranearecalled
phosphoglycerides
Aphosphoglyceride consistsofphosphoricacidandfattyacidscalledglycerides
Theheadofthismolecule,thephosphoglyceride,ishydrophilic (attractedto
water)
Thefattyacidshavetailsconsistingofhydrocarbonchainswhicharehydrophobic
(repelledbywater)
Iffattyacidmoleculesareplacedinwater,theyformlittleclumps,withtheacid
headsthatareattractedtowaterontheoutside,andthehydrocarbontailsthat
arerepelledbywaterontheinside.
TheCellMembrane
ad
e
ng
ho
.G
r.R
.B
Ifthesemoleculesareverycarefullyplacedonawatersurface,theyorient
themselvessothatallacidheadsareinthewaterandallhydrocarbontails
protrudefromit.
Ifanotherlayerofmoleculeswereaddedandmorewaterputontop,the
hydrocarbontailswouldlineupwiththosefromthefirstlayer,toformadouble
(twomoleculesthick)layer.
Theacidheadswouldprotrudeintothewateroneachsideandthehydrocarbons
wouldfillthespacebetween.
Thisbilayeristhebasicstructureofthecellmembrane.
Fromthebioelectricviewpoint,theionicchannels constituteanimportantpartof
thecellmembrane
Thesearemacromolecularporesthroughwhichsodium,potassium,andchloride
ionsflowthroughthemembrane.
Theflowoftheseionsformsthebasisofbioelectricphenomena.
TheCellMembrane
ad
e
ng
ho
.G
r.R
.B
D
Theconstructionofacellmembrane
Themainconstituentsaretwolipidlayers,withthehydrophobictailspointinginsidethe
membrane(awayfromtheaqueousintracellularandinterstitialmediums).
Themacromolecularporesinthecellmembraneformtheionicchannelsthroughwhich
sodium,potassium,andchloridemoleculesflowthroughthemembraneandgeneratethe
bioelectricphenomena
TheSynapse
ad
e
r.R
.B
.G
ng
The junction between an axon and the next cell with which it communicates is called the
synapse.
Information proceeds from the cell body unidirectionally over the synapse, first along
the axon and then across the synapse to the next nerve or muscle cell.
The part of the synapse that is on the side of the axon is called the presynaptic terminal;
that part on the side of the adjacent cell is called the postsynaptic terminal
ho
Thesynapsebetweenamotornerveandthemuscleitinnervatesiscalledthe
neuromuscularjunction
MuscleCell
ad
e
Threetypesofmusclesinthebody
r.R
.B
.G
ho
ng
smoothmuscle
striatedmuscle(skeletalmuscle)
cardiacmuscle
Smoothmuscles
Theyareinvoluntary(i.e.,theycannotbecontrolledvoluntarily)
cellshaveavariablelengthbutareintheorderof0.1mmexist,for
example,inthedigestivetract,inthewallofthetrachea,uterus,
andbladder
Thecontractionofsmoothmuscleiscontrolledfromthebrain
throughtheautonomicnervoussystem
Striatedmuscles
Cardiacmuscle
r.R
.B
.G
ho
ng
ad
e
alsocalledskeletalmuscles becauseoftheiranatomicallocation,areformedfroma
largenumberofmusclefibers,thatrangeinlengthfrom1to40mmandindiameter
from0.01to0.1mm.
Eachfiberformsa(muscle)cellandisdistinguishedbythepresenceofalternatingdark
andlightbands.
Thestriatedmusclefibercorrespondstoan(unmyelinated)nervefiberbutis
distinguishedelectrophysiologically fromnervebythepresenceofaperiodictransverse
tubularsystem(TTS),acomplexstructurethat,ineffect,continuesthesurface
membraneintotheinteriorofthemuscle.
Propagationoftheimpulseoverthesurfacemembranecontinuesradiallyintothefiber
viatheTTS,andformsthetriggerofmyofibrillar contraction.
ThepresenceoftheTTSaffectsconductionofthemusclefibersothatitdiffers
(althoughonlyslightly)frompropagationonan(unmyelinated)nervefiber.
Striatedmusclesareconnectedtothebonesviatendons.
Suchmusclesarevoluntaryandformanessentialpartoftheorganofsupportand
motion.
alsostriated,butdiffersinotherwaysfromskeletalmuscle
Notonlyisitinvoluntary,butalsowhenexcited,itgeneratesamuchlongerelectric
impulsethandoesskeletalmuscle,lastingabout300ms
Correspondingly,themechanicalcontractionalsolastslonger
cardiacmusclehasaspecialproperty:Theelectricactivityofonemusclecellspreadsto
allothersurroundingmusclecells,owingtoanelaboratesystemofintercellular
junctions.
r.R
.B
.G
ho
ng
ad
e
StructureofMuscleCell
Biopotentials
ng
ad
e
Certainsystemsofthebodycreatetheirown"monitoring"
signals,whichconveyusefulinformationregardingthe
functionstheyrepresent.
r.R
.B
.G
ho
ThesesignalsaretheBiopotentials BPassociatedwiththe
conductionalongthesensoryandmotornervoussystem,
muscularcontractions,brainactivity,heartcontractions,etc.
Thesepotentialsarearesultoftheelectrochemicalactivity
occurringincertainclassesofcellswithinthebody
ExcitableCells.
MeasurementsoftheseBiopotentialscanprovideclinicians
withinvaluablediagnosticinformation
Biopotentials
ad
e
ng
ho
.G
r.R
.B
Fromthebiologicalcell,electricalpotentialsaregeneratedduetothe
electrolytesinsideandoutsideofthecell
Abioelectricpotentialmaybedefinedasthedifferenceinpotential
betweentheinsideandtheoutsideofacell;thereexistsadifferencein
potentialexistingacrossthecellwallormembrane.
Acellconsistsofanionicconductorseparatedfromtheoutside
environmentbyasemipermeableorselectivelypermeablecell
membrane
Humancellsmayvaryfrom1micronto100micronsindiameter,from1
millimeterto1meterinlengthandhaveatypicalmembranethicknessof
100Angstromunits
Bioelectricityisstudiedbothfromtheviewpointofthesourceofelectrical
energywithinthecellandalsofromtheviewpointofthelawsof
electrolyticcurrentflowrelativetotheremoteionicfieldsproduced
currentsbythecell.
Wemakemeasurementsexternaltoagroupofcellswhilethesecellsare
supplyingelectrolyticcurrentflow.
CellPotentialGenesis
r.R
.B
.G
ho
ng
ad
e
Experimentalinvestigationswithmicroelectrodeshaveshown
thattheinternalrestingpotentialwithinacellis60mVto90
mV(typically 70mV)withreferencetotheoutsideofthecell
Byconvention,theoutsideisdefinedas0mV(ground)
Thispotentialchangestoapproximately+20mVforashort
periodduringcelldepolarisation
Cellactivityresultsfromsomeformofstimulation
CellMembranePotentials
ad
e
r.R
.B
ng
ho
Cellmembranes ingeneral,andmembranesofnervecellsinparticular,
maintainasmallvoltageor"potential"acrossthemembraneinitsnormal
orrestingstate.
Inthereststate,theinsideofthenervecellmembraneisnegativewith
respecttotheoutside(typicallyabout70millivolts).
Thevoltagearisesfromdifferencesinconcentration oftheelectrolyte
ionsK+andNa+.
ThereisaprocesswhichutilizesATP(adenosinetriphosphate Active
transportofionsagainst anestablishedelectrochemicalgradient )to
pumpoutthreeNa+ionsandpumpintwoK+ions.Thecollectiveactionof
thesemechanismsleavestheinteriorofthemembraneabout70mVwith
respecttotheoutside.
Iftheequilibriumofthenervecellisdisturbedbythearrivalofasuitable
stimulus dynamicchanges inthemembranepotentialinresponseto
thestimulusiscalledanActionPotential.
Aftertheactionpotentialthemechanismsdescribedabovebringthecell
membranebacktoitsrestingstate
.G
ExcitableCells
ng
ad
e
Excitablecellsareaclassofcellsthatproducebioelectric
potentialsasaresultofelectrochemicalactivity.
.G
ho
Atanygiventime,thesecellscanexistinoneoftwostates,
resting andactive.
r.R
.B
Chemicalandelectricalstimulicanforceanexcitablecellfrom
therestingtotheactivestate.
Whiletherearenumerousionicspeciespresentbothinside
andoutsidethecell,onlythreeions(forwhichthecell
membraneinitsrestingstateispermeable) playakeyrolein
thebehaviorofthesecells:K+,Na+ andCl.
RestingPotential,IonicConcentrations,andChannels
ad
e
ng
ho
.G
r.R
.B
Theneuroncellmembraneisapproximately10nmthickand,becauseitconsistsofalipid
bilayer(i.e.,twoplatesseparatedbyaninsulator),hascapacitiveproperties.
TheextracellularfluidiscomposedofprimarilyNa+ andCl,andtheintracellularfluid
(cytoplasm)iscomposedofprimarilyK+ andA
Thelargeorganicanions(A)areprimarilyaminoacidsandproteinsanddonotcrossthe
membrane.
Almostwithoutexception,ionscannotpassthroughthecellmembraneexceptthrougha
channel
Channelsallowionstopassthroughthemembrane,areselective,andare
eitherpassiveoractive
Passivechannelsarealwaysopenandareionspecific
Aparticularchannelallowsonlyoneiontypetopassthroughthe
membraneandpreventsallotherionsfromcrossingthemembrane
throughthatchannel.
PassivechannelsexistforCl,K+,andNa+
r.R
.B
.G
ho
ng
ad
e
r.R
.B
.G
ho
ng
ad
e
Activechannels,orgates,areeitheropenedorclosedin
responsetoanexternalelectricalorchemicalstimulation.
Theactivechannelsarealsoselectiveandallowonlyspecific
ionstopassthroughthemembrane.
Typically,activegatesopeninresponsetoneurotransmitters
andanappropriatechangeinmembranepotential.
ActiveState
r.R
.B
.G
ho
ng
ad
e
Ifadequatelystimulated,eitherelectricallyor
chemically,theexcitablecellwillenterintotheactive
state.
Thetransmembranepotentialvarieswithtimeand
positionwithinthecellinthisstate,andiscalledan
actionpotential.
Thefollowingsequenceofeventsoccurswhenthe
cellenterstheactivestate:
Thechemicalorelectricalstimuliincreasesthe
permeabilityofthemembranetoNa+
Na+ rushesintothecellduetothelargeconcentration
gradient.
ActiveState(cont.)
r.R
.B
.G
ho
ng
ad
e
Thesepositivelychargedionsenteringthecellcausethe
transmembranepotentialtobecomelessnegative,and
eventuallyslightlypositive
Thischangeisoftenreferredtoasadepolarization
Ashorttime(tenthsofmicroseconds)laterthemembranes
permeabilitytoK + increases,whichresultsinanoutflowofK+
TheoutflowofK + causesthetransmembranepotentialto
decrease
Thisdecreaseinpotentialcausesthemembranes
permeabilitytobothNa +,andeventuallyK +,todecreaseto
theirrestinglevels
Thereisonlyarelativelysmall(immeasurable)netflowof
ionsacrossthemembraneduringanactionpotential.
TheNaKpumprestorestheconcentrations(pumpsNaout
andKin)oftheionstotheirrestinglevels.
Theresultofthetransitionfromtherestingtothe
activestateistheActionPotential
ho
ng
ad
e
Inresponsetotheappropriatestimulus,thecell
membraneofanervecellgoesthroughasequenceof
depolarizationfromitsreststatetotheactivestate
followedbyrepolarizationtothereststateonceagain
r.R
.B
.G
Thecellmembraneactuallyreversesitsnormalpolarity
forabriefperiodbeforereestablishingtherest
potential
Theactionpotentialsequenceisessentialforneural
communication.
Thesimplestactioninresponsetothoughtrequires
manysuchactionpotentialsforitscommunicationand
performance
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
Theprocesssummary
5.
6.
ad
e
r.R
.B
4.
3.
.G
ho
2.
Astimulus isreceivedbythedendritesofanervecell.ThiscausestheNa+
channelstoopen.Iftheopeningissufficienttodrivetheinteriorpotential
from70mVupto55mV,theprocesscontinues.
Havingreachedtheactionthreshold,moreNa+channels(sometimescalled
voltagegatedchannels)open TheNa+influxdrivestheinteriorofthecell
membraneuptoabout+30mV.Theprocesstothispointiscalled
DEPOLARIZATION.
TheNa+channelscloseandtheK+channelsopen.HavingbothNa+andK+
channelsopenatthesametimewoulddrivethesystemtowardneutrality
andpreventthecreationoftheactionpotential.
WiththeK+channelsopen,themembranebeginstoREPOLARIZE back
towarditsrestpotential.
Therepolarizationtypicallyovershootstherestpotentialtoabout90mV.
Thisiscalledhyperpolarization. Hyperpolarization preventstheneuronfrom
receivinganotherstimulusduringthistime.
Afterhyperpolarization,theNa+/K+pumpseventuallybringthemembrane
backtoitsrestingstate of70mV.
ng
1.
ad
e
ng
ho
.G
r.R
.B
Absolute&RelativeRefractoryPeriod
ARP&RRP
ho
ng
ad
e
DuringtheinitialportionoftheActionpotential
membranedoesnotrespond Absoluterefractory
period
r.R
.B
.G
DuringtheRelativeRefractoryPeriod RRPtheaction
potentialtakesaction
Therefractoryperiod limitsthefrequencyofa
repetitiveexcitationprocedure
e.g.ARP=1ms
upperlimitofrepetitivedischarge
<1000impulses/s
Absolute&RelativeRefractoryPeriod
ARP&RRP(cont.)
ad
e
Nernstequil.PotforNa
r.R
.B
.G
ho
ng
v:actionpot.
Nernstequil.PotforK
r.R
.B
.G
ho
ng
ad
e
ElectricalCircuitModelofNerveMembrane
r.R
.B
.G
ho
ng
ad
e
Bioelectricphenomena
Lecture3
r.R
.B
.G
ho
ng
ad
e
Biotransducers
PartI:BioElectrodes
PartII:PhysicalMeasurements
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
Introduction
r.R
.B
.G
ho
ng
ad
e
Biomedicalsensorsareusedroutinelyinclinical
medicineandbiologicalresearchformeasuringa
widerangeofphysiologicalvariables
Oftencalledbiomedicaltransducersandarethe
mainbuildingblocksofdiagnosticmedical
instrumentation
Usedinvivotoperformcontinuousinvasiveand
noninvasivemonitoringofcriticalphysiological
variables
Alsousedinvitrotohelpcliniciansinvarious
diagnosticprocedures
r.R
.B
.G
ho
ng
ad
e
Somesensorsareusedprimarilyinclinical
laboratoriestomeasureinvitrophysiological
quantitiessuchaselectrolytes,enzymes,and
otherbiochemicalmetabolitesinblood
Otherbiomedicalsensorsformeasuring
pressure,flow,andtheconcentrationsof
gasessuchasoxygenandcarbondioxideare
usedinvivotofollowcontinuously(monitor)
theconditionofapatient
RequirementsofBiomedicalSensors
ng
ho
.G
accuracy,
operatingrange
responsetime
Sensitivity
Resolution
Reproducibility
r.R
.B
ad
e
Stringentrequirements
Assessinvitrothe
Later,dependingontheintendedapplication,similarin
vivotestsmayberequiredtoconfirmthespecificationsof
thesensorandtoassurethatthemeasurementremains
Sensitive
Stable
Safe
costeffective
SensorClassifications
ng
ho
.G
Geometric
Mechanical
Thermal
Hydraulic
Electric
Optical
r.R
.B
ad
e
Physicalsensors
Chemicalsensors
Gas
Electrochemical
Photometric
Otherphysicalchemicalmethods
Bioanalytic
SensorClassifications
ho
ng
ad
e
Classifiedaccordingtothequantitytobemeasuredandare
typicallycategorizedasphysical,electrical,orchemical
dependingontheirspecificapplications
Biosensors areaspecialsubclassificationofbiomedical
sensors
Theyhavetwodistinctcomponents:
r.R
.B
.G
abiologicalrecognitionelementsuchasapurifiedenzyme,antibody,
orreceptor,(whichfunctionsasamediatorandprovidestheselectivity
thatisneededtosensethechemicalcomponent (usuallyreferredtoas
theanalyte)ofinterest)
asupportingstructure,whichalsoactsasatransducerandisin
intimatecontactwiththebiologicalcomponent.(Thepurposeofthe
transduceristoconvertthebiochemicalreactionintotheformofan
optical,electrical,orphysicalsignalthatisproportionaltothe
concentrationofaspecificchemical)
Anotherwayofclassifyingbiomedicaltransducers!
r.R
.B
.G
ho
ng
ad
e
Onecanalsolookatbiomedicalsensorsfromthestandpoint
oftheirapplications
Thesecanbegenerallydividedaccordingtowhetherasensor
isusedfor
diagnostic
therapeutic*purposes
Sensorsforclinicalstudiessuchasthosecarriedoutinthe
clinicalchemistrylaboratorymustbestandardizedinsucha
waythaterrorsthatcouldresultinanincorrectdiagnosisor
inappropriatetherapyarekepttoanabsoluteminimum
Thesesensorsmustnotonlybereliablethemselves,but
appropriatemethodsmustexistfortestingthesensorsthat
areapartoftheroutineuseofthesensorsformaking
biomedicalmeasurements
*Havingorexhibitinghealingpowers
Onemorewayofclassifyingbiomedicaltransducers!
ng
ad
e
Standpointofhowtheyareappliedtothe
patientorresearchsubject
r.R
.B
.G
ho
Noncontacting(noninvasive)
Skinsurface(contacting)
Indwelling(minimallyinvasive)
Implantable(invasive)
ad
e
ng
ho
r.R
.B
.G
PartI:BioElectrodes
BIOPOTENTIALMEASUREMENTS
r.R
.B
.G
ho
ng
ad
e
Biopotential measurementsaremadeusing
differentkindsofspecializedelectrodes
Thefunctionoftheseelectrodesistocouple
theionicpotentialsgeneratedinsidethebody
toanelectronicinstrument
Biopotential electrodesareclassifiedeitheras
noninvasive(skinsurface)
invasive(e.g.,microelectrodesorwireelectrodes)
r.R
.B
.G
ho
ng
ad
e
BioelectricSignalsSensedbyBiopotential Electrodes
andTheirSources
TheElectrolyte/MetalElectrode
Interface
r.R
.B
.G
ho
ng
ad
e
r.R
.B
.G
ho
ng
ad
e
Thislocalizedchargedistributioncausesanelectricpotential,
calledahalfcellpotential,tobedevelopedacrossthe
interfacebetweenthemetalandtheelectrolytesolution
r.R
.B
.G
ad
e
ng
Biopotential measurementsaremadebyutilizingtwosimilarelectrodes
composedofthesamemetal.
Therefore,thetwohalfcellpotentialsfortheseelectrodeswouldbeequal
inmagnitude.
Forexample,twosimilarbiopotential electrodescanbetapedtothechest
nearthehearttomeasuretheelectricalpotentialsgeneratedbytheheart
(electrocardiogram,orECG)
Ideally,assumingthattheskintoelectrodeinterfacesareelectrically
identical,thedifferentialamplifierattachedtothesetwoelectrodeswould
amplifythebiopotential (ECG)signalbutthehalfcellpotentialswouldbe
canceledout
Inpracticedisparityinelectrodematerialorskincontactresistancecould
causeasignificantDCoffsetvoltagethatwouldcauseacurrenttoflow
throughthetwoelectrodes.
Thiscurrentwillproduceavoltagedropacrossthebody
Theoffsetvoltagewillappearsuperimposedattheoutputoftheamplifier
andmaycauseinstabilityorbaselinedriftintherecordedbiopotential
ho
ad
e
a)
ng
b)
r.R
.B
.G
ho
a)Ifelectrodehassamematerialascation,thenthismaterialgetsoxidizedandenters
theelectrolyteasacationandelectronsremainattheelectrodeandflowinthe
externalcircuit.
b)Ifanioncanbeoxidizedattheelectrodetoformaneutralatom,oneortwoelectrons
aregiventotheelectrode.
Thedominatingreactioncanbeinferredfromthefollowing:
Currentflowfromelectrodetoelectrolyte:Oxidation (Lossofe)
Currentflowfromelectrolytetoelectrode:Reduction (Gainofe)
ng
ad
e
Acharacteristicpotentialdifferenceestablishedbytheelectrodeandits
surroundingelectrolytewhichdependsonthemetal,concentrationofions
insolutionandtemperature(andsomesecondorderfactors).
.G
ho
Halfcellpotentialcannotbemeasuredwithoutasecondelectrode.
r.R
.B
Thehalfcellpotentialofthestandardhydrogenelectrodehasbeen
arbitrarilysettozero.Otherhalfcellpotentialsareexpressedasapotential
differencewiththiselectrode.
ReasonforHalfCellPotential:ChargeSeparationatInterface
Oxidationorreductionreactionsattheelectrodeelectrolyteinterfaceleadtoadouble
chargelayer,similartothatwhichexistsalongelectricallyactivebiologicalcellmembranes.
r.R
.B
.G
ho
ng
ad
e
Note:Electrodematerialismetal+saltorpolymerselectivemembrane
Polarization
ng
ad
e
Ifthereisacurrentbetweentheelectrodeandelectrolyte,theobservedhalfcell
potentialisoftenalteredduetopolarization.
Resistance
Currentchangesresistance
ofelectrolyteandthus,
avoltagedropresults.
r.R
.B
.G
ho
Overpotential
Differencebetweenobserved
andzerocurrenthalfcell
potentials
Concentration
Changesindistribution
ofionsattheelectrode
electrolyteinterface
Activation
Theactivationenergy
barrierdependsonthe
directionofcurrentand
determineskinetics
Note:Polarizationandimpedanceoftheelectrodearetwoofthemostimportant
electrodepropertiestoconsider.
Nernst Equation
ho
Forthegeneraloxidationreductionreaction
ng
ad
e
Whentwoaqueousionicsolutionsofdifferentconcentrationareseparatedbyanion
selectivesemipermeablemembrane,anelectricpotentialexistsacrossthemembrane.
r.R
.B
.G
Note:interestedin
ionicactivityatthe
electrode
(butnotetemp
dependence
TheNernstequationforhalfcellpotentialis
whereE0 :StandardHalfCellPotential
E:HalfCellPotential
a:IonicActivity(generallysameasconcentration)
n:Numberofvalenceelectronsinvolved
ad
e
Useforrecording
r.R
.B
PerfectlyNonPolarizableElectrode
.G
ho
ng
Theseareelectrodesinwhichnoactualchargecrossestheelectrodeelectrolyteinterface
whenacurrentisapplied.Thecurrentacrosstheinterfaceisadisplacementcurrentandthe
electrodebehaveslikeacapacitor.Example:Ag/AgClElectrode
Theseareelectrodeswherecurrentpassesfreelyacrosstheelectrodeelectrolyteinterface,
requiringnoenergytomakethetransition.Theseelectrodesseenooverpotentials.Example
:Platinumelectrode
Example:AgAgClisusedinrecordingwhilePtisuseinstimulation
Useforstimulation
Ag/AgCl Electrode
ho
ng
ad
e
Relevantionicequations
r.R
.B
GoverningNernstEquation
Ag+Cl
.G
Cl2
Solubility
productofAgCl
ad
e
Equivalent Circuit
.G
ho
ng
Cd :capacitanceofelectrodeeletrolyteinterface
Rd :resistanceofelectrodeeletrolyteinterface
Rs:resistanceofelectrodeleadwire
Ecell :cellpotentialforelectrode
r.R
.B
Rd+Rs
Cornerfrequency
Rs
FrequencyResponse
Rd
ho
Sweatglands
andducts
Rs
Gel
.G
100
r.R
.B
Ese
StratumCorneum
Epidermis
Nerve
endings
Re
Dermisand
subcutaneouslayer
100
Ce
Capillary
ng
Electrode
ad
e
Ehe
EP
CP
RP
Alterskin
transport(or
deliverdrugs)
by:
Pores
producedby
laser,
ultrasoundor
by
iontophoresis
Ru
Skinimpedancefor1cm2patch:
200k @1Hz
200 @1MHz
Motion Artifact
ad
e
Why
.G
ho
ng
Whentheelectrodemoveswithrespecttotheelectrolyte,thedistributionofthe
doublelayerofchargeonpolarizableelectrodeinterfacechanges.Thischangesthehalf
cellpotentialtemporarily.
r.R
.B
What
Ifapairofelectrodesisinanelectrolyteandonemoveswithrespecttotheother,
apotentialdifferenceappearsacrosstheelectrodesknownasthemotionartifact.
Thisisasourceofnoiseandinterferenceinbiopotentialmeasurements
Motionartifactisminimalfornonpolarizableelectrodes
.G
ho
ng
ad
e
Electrolyte
r.R
.B
(historic interest)
2. Suction Electrodes
3. Floating Electrodes
4. Flexible Electrodes
Thinkofthe
constructionof
electrosurgical
electrode
And,howdoes
electrosurgery
work?
ng
r.R
.B
.G
ho
ad
e
(a)Metalplateelectrodeusedforapplicationtolimbs.
(b)Metaldiskelectrodeappliedwithsurgicaltape.
(c)Disposablefoampadelectrodes,oftenusedwithECG
ng
ho
r.R
.B
.G
Suctionelectrodes
Nostrapsoradhesivesrequired
precordial(chest)ECG
canonlybeusedforshortperiods
ad
e
CommonlyUsedBiopotential
Electrodes
Floatingelectrodes
metaldiskisrecessed
swimmingintheelectrolytegel
notincontactwiththeskin
reducesmotionartifact
SuctionElectrode
Insulating
package
.G
ho
ng
Doublesided
Adhesivetape
ring
ad
e
CommonlyUsedBiopotential
Electrodes Metaldisk
(a)
r.R
.B
SnapcoatedwithAgAgCl
(b)
Externalsnap
Gelcoatedsponge
Plasticdisk
Reusable
Disposable
Plasticcup
Electrolytegel
inrecess
Foampad
Deadcellularmaterial
Tack
Capillaryloops Germinatinglayer
(c)
FloatingElectrodes
ng
ho
r.R
.B
.G
Flexibleelectrodes
Bodycontoursareoftenirregular
Regularlyshapedrigidelectrodes
maynotalwayswork.
Specialcase:infants
Material:
Polymerornylonwithsilver
Carbonfilledsiliconrubber
(Mylarfilm)
ad
e
(a)Carbonfilledsiliconerubberelectrode.
(b)Flexiblethinfilmneonatalelectrode.
(c)Crosssectionalviewofthethinfilm
electrodein(b).
ad
e
Internal Electrodes
ho
.G
r.R
.B
(a)Insulatedneedleelectrode.
(b)Coaxialneedleelectrode.
(c)Bipolarcoaxialelectrode.
(d)Finewireelectrodeconnected
tohypodermicneedle,before
beinginserted.
(e)Crosssectionalviewofskin
andmuscle,showingcoiled
finewireelectrodeinplace.
ng
Needleandwireelectrodesfor
percutaneousmeasurementof
biopotentials
r.R
.B
.G
ho
ng
ad
e
Biopotential microelectrodes:
(a)capillaryglassmicroelectrode
(b)insulatedmetalmicroelectrode
(c)solidstatemultisiterecordingmicroelectrode
r.R
.B
.G
ho
ng
ad
e
Electrodesfordetectingfetalelectrocardiogramduringlabor,bymeans
ofintracutaneousneedles(a)Suctionelectrode.(b)Crosssectionalviewof
suctionelectrodeinplace,showingpenetrationofprobethroughepidermis.
(c)Helicalelectrode,whichisattachedtofetalskinbycorkscrewtypeaction.
Electrode Arrays
Insulated leads
ad
e
Contacts
ho
(a)
Tines
r.R
.B
Exposed tip
Base
(c)
Ag/AgCl electrodes
ng
Base
Insulated leads
.G
Contacts
Ag/AgCl electrodes
Base
(b)
Examplesofmicrofabricatedelectrodearrays.
(a)Onedimensionalplungeelectrodearray,
(b)Twodimensionalarray,and
(c)Threedimensionalarray
Microelectrodes
ng
ad
e
Why
Measure potential difference across cell membrane
r.R
.B
.G
ho
Requirements
Small enough to be placed into cell
Intracellular
Strong enough to penetrate cell membrane
Extracellular
Typical tip diameter: 0.05 10 microns
Types
Solid metal -> Tungsten microelectrodes
Supported metal (metal contained within/outside glass needle)
Glass micropipette -> with Ag-AgCl electrode metal
Metal Microelectrodes
Microns!
r.R
.B
.G
ho
ng
ad
e
Extracellularrecording typicallyinbrainwhereyouareinterestedin
recordingthefiringofneurons(spikes).
Usemetalelectrode+insulation>goestohighimpedance
amplifiernegativecapacitanceamplifier!
r.R
.B
.G
ho
ng
ad
e
(a)Metalinsideglass
(b)Glassinsidemetal
Glass Micropipette
heat
ng
ad
e
pull
AgAgClwire+3M
KClhasverylow
junctionpotential
andhencevery
accuratefordc
measurements
(e.g.action
potential)
r.R
.B
.G
ho
Aglassmicropipetelectrodefilled
withanelectrolyticsolution
(a)Sectionoffineboreglass
capillary.
(b)Capillarynarrowedthrough
heatingandstretching.
(c)Finalstructureofglasspipet
microelectrode.
Fillwith
intracellularfluid
or3MKCl
Intracellularrecording typicallyforrecordingfromcells,suchascardiacmyocyte
Needhighimpedanceamplifiernegativecapacitanceamplifier!
ad
e
Electrical Properties of
Microelectrodes
r.R
.B
.G
ho
ng
MetalMicroelectrode
Metalmicroelectrodewithtipplaced
withincell
Usemetalelectrode+insulation>goestohighimpedance
amplifiernegativecapacitanceamplifier!
Equivalentcircuits
ElectricalPropertiesofGlassIntracellular
Microelectrodes
r.R
.B
.G
ho
ng
ad
e
GlassMicropipetteMicroelectrode
Stimulating Electrodes
ad
e
Features
r.R
.B
.G
ho
ng
Cannotbemodeledasaseriesresistanceandcapacitance
(thereisnosingleusefulmodel)
Thebody/electrodehasahighlynonlinearresponseto
stimulation
Largecurrentscancause
Platinumelectrodes:
Cavitation
Applications:neuralstimulation
Celldamage
Heating
ModerndayPtIrandotherexotic
metalcombinationstoreduce
polarization,improveconductance
andlonglife/biocompatibility
Steelelectrodesforpacemakersand
defibrillators
r.R
.B
.G
ho
ng
ad
e
IntraocularStimulationElectrodes
Microelectronic technology
for Microelectrodes
Bondingpads
Sisubstrate
.G
Exposedtips
Siliconprobe
ho
ng
Exposed
electrodes
ad
e
Insulated
leadvias
SiO2 insulated
Auprobes
Channels
Siliconchip
Leadvia
Hole
(b) Multielectrodesiliconprobe
Miniature
insulating
chamber
r.R
.B
(a) Beamleadmultipleelectrode.
Siliconprobe
Electrode
(c) Multiplechamberelectrode
Contact
metalfilm
(d)
Peripheralnerveelectrode
Differenttypesofmicroelectrodesfabricatedusingmicrofabrication/MEMS
technology
.G
ho
ng
ad
e
Dissimilarmetalshavedifferenthalfcellpotentialsmakingan
electricallyunstable,noisyjunction.
Iftheleadwireisadifferentmetal,besurethatitiswellinsulated.
Donotletasolderjunctiontouchtheelectrolyte.Ifthejunctionmust
touchtheelectrolyte,fabricatethejunctionbyweldingormechanical
clampingorcrimping.
r.R
.B
Fordifferentialmeasurements,usethesamematerialfor
eachelectrode.
Ifthehalfcellpotentialsarenearlyequal,theywillcancelandminimize
thesaturationeffectsofhighgain,dccoupledamplifiers.
Electrodesattachedtotheskinfrequentlyfalloff.
Useveryflexibleleadwiresarrangedinamannertominimizetheforce
exertedontheelectrode.
Tapetheflexiblewiretotheskinashortdistancefromtheelectrode,
makingthisastressreliefpoint.
Matchtheleadwireinsulationtothespecificapplication.
.G
ho
ng
ad
e
Repeatedflexingcanbreakthewireinsideitsinsulation.
Provestrainreliefbycreatingagradualmechanicaltransitionbetweenthewire
andtheelectrode.
Useataperedregionofinsulationthatgraduallyincreasesindiameterfromthat
ofthewiretowardsthatoftheelectrodeasonegetscloserandclosertothe
electrode.
r.R
.B
Iftheleadwiresandtheirjunctionstotheelectrodearesoakedinextracellular
fluidoracleaningsolutionforlongperiodsoftime,waterandothersolventscan
penetratethepolymericcoatingandreducetheeffectiveresistance,makingthe
leadwirebecomepartoftheelectrode.
Suchanelectrodecapturesothersignalsintroducingunwantednoise.
Matchyouramplifierdesigntothesignalsource.
Besurethatyouramplifiercircuithasaninputimpedancethatismuchgreater
thanthesourceimpedanceoftheelectrodes.
ad
e
ng
ho
r.R
.B
.G
PartII:PhysicalMeasurements
ConventionalTransducers
ad
e
PrimaryTransducers
large,butgenerallyreliable,basedonoldertechnology
.G
MicroelectronicSensors
ho
ng
thermocouple:temperaturedifference
compass(magnetic):direction
millimetersized,highlysensitive,lessrobust
r.R
.B
photodiode/phototransistor:photonenergy(light)
infrareddetectors,proximity/intrusionalarms
piezoresisitve pressuresensor:air/fluidpressure
microaccelerometers:vibration,velocity(carcrash)
chemicalsensors:O2,CO2,Cl,Nitrates(explosives)
DNAarrays:matchDNAsequences
Directvs.IndirectMeasurement
ad
e
DirectMeasurement:
.G
ho
ng
Whensensordirectlymeasuresparameterofinterest
Example,displacementsensormeasuringdiameterof
bloodvessel
r.R
.B
IndirectMeasurement:
Whensensormeasuresaparameterthatcanbe
translatedintotheparameterofinterest
Example,displacementsensormeasuringmovement
ofamicrophonediaphragmtoquantifyblood
movementthroughtheheart
r.R
.B
.G
ho
ng
ad
e
PhysicalVariablesandSensors
DisplacementMeasurements
ng
ho
requiredisplacementsensors
ad
e
Manybiomedicalparametersrelyonmeasurementsofsize,
shape,andpositionoforgans,tissue,etc.
.G
Examples(direct)diameterofbloodvessel
r.R
.B
(indirect)movementofamicrophonediaphragmtoquantify
bloodmovementthroughtheheart
ResistiveSensors(Potentiometers&StrainGages)
InductiveSensors
CapacitiveSensors
PiezoelectricSensors
PrimaryTransducerTypes
r.R
.B
.G
ho
ng
ad
e
DisplacementSensors
VariableResistanceSensor(Potentiometer)
r.R
.B
.G
ho
ng
Potentiometersproduceoutputpotential
(voltage)changeinresponsetoinput
(e.g.,displacement)changes
typicallyformedwithresistive
elementse.g.carbon/metalfilm
V=IR
producelinearoutputinresponseto
displacement
Examplepotentiometricdisplacement
sensors
Translational:small(~mm)linear
displacements
Vo increasesasxi increases
SingleTurn:small(1050)rotational
displacements
Vo increasesas increases
ad
e
Apotentiometerisaresistivetypetransducerthatconvertseither
linearorangulardisplacementintoanoutputvoltagebymovinga
slidingcontactalongthesurfaceofaresistiveelement
xi
VariableResistanceSensor(StrainGauge)
Straingaugesaredisplacementtypetransducersthatmeasurechangesin
thelengthofanobjectasaresultofanappliedforce
Thesetransducersproducearesistancechangethatisproportionaltothe
fractionalchangeinthelengthoftheobject,alsocalledstrain
Therelativesensitivityofthisdeviceisgivenbyitsgaugefactor,
ho
ng
ad
e
r.R
.B
.G
WhereRisthechangeinresistancewhenthe
structureisstretchedbyanamountl
r.R
.B
.G
ho
ng
ad
e
(a)Bondedtypestrain
gaugetransducer
(b) Resistivestraingauge
(unbondedtype)blood
pressuretransducer
Straingaugesonacantileverstructuretoprovide
temperaturecompensation
r.R
.B
.G
ho
ng
ad
e
(a)crosssectional
viewofthecantilever
(b)placementofthestraingaugesinahalfbridgeorfullbridgefor
temperaturecompensationandenhancedsensitivity
Liquidmetalstraingauge
Insteadofusingasolidelectricconductorsuchasthewireormetalfoil,
mercuryconfinedtoacompliant,thinwall,narrowboreelastomerictube
isused
Thecomplianceofthisstraingaugeisdeterminedbytheelastic
propertiesofthetube
Sinceonlytheelasticlimitofthetubeisofconcern,thissensorcanbe
usedtodetectmuchlargerdisplacementsthanconventionalstrain
gauges
Itssensitivityisroughlythesameasafoilorwirestraingauge,butitis
notasreliable
r.R
.B
.G
ho
ng
ad
e
Elasticresistancestraingagesareextensivelyusedin
biomedicalapplications,especiallyincardiovascularand
respiratorydimensionalandplethysmographic (volume
measuring)determinations
Elasticstraingageistypicallylinearwith1%for10%of
maximalextensionthus,straingagesareonlygoodmeasuring
smalldisplacements
r.R
.B
.G
ho
ng
ad
e
Mercuryinrubberstraingageplethysmography
(a) Fourleadgageappliedtohumancalf
(b) Bridgeoutputforvenousocclusionplethysmography
(c)Bridgeoutputforarterialpulseplethysmography
Semiconductorstraingauge
r.R
.B
.G
ho
ng
ad
e
Thesedevicesarefrequentlymadeoutofpieces
ofsiliconwithstraingaugepatternsformedusing
semiconductormicroelectronictechnology.
Theprincipaladvantageofthesedevicesisthat
theirgaugefactorscanbemorethan50times
greaterthanthatofthesolidandliquidmetal
devices
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ho
ng
ad
e
StrainGauge:Materials
Gforsemiconductormaterials~5070xthatofmetalsduetostrongerpiezo
resistiveeffect
semiconductorshavemuchhigherTCR,requirestemperaturecompensationin
straingauge
r.R
.B
.G
ho
ng
ad
e
Disposablebloodpressuresensor
Madeofclearplasticsothatairbubblescanbeseen
SalineflowsfromintravenousbagthroughclearIVtubeandthesensorto
thepatient
Thisflushesbloodoutofthetipofthecathetertopreventclotting
Alevercanopenorclosetheflushvalve
ThesiliconchiphassilicondiaphragmwithfourresistorWheatstonebridge
diffusedtoit
Itsisisolatedelectricallybyasiliconeelastomergel
InductiveSensors
r.R
.B
.G
ho
ng
ad
e
AninductanceLcanbeusedtomeasure
displacementbyvaryinganythreeofthecoil
parameters:
where
n=numberofturnsofcoil
G=geometricformfactor
=effectivepermeabilityofthe
medium
Eachoftheseparameterscanbechangedbymechanical
means
r.R
.B
.G
ho
ng
ad
e
Inductivedisplacementsensors
(a)selfinductance,(b)mutualinductance,(c)differentialtransformer
ng
ho
.G
r.R
.B
The
linear
variable
differential
transformer (LVDT) is widely used in
physiological research and clinical
medicine to measure pressure,
displacement, and force
The LVDT is composed of a primary coil
and two secondary coils connected in
series
The coupling between these two coils is
changed by the motion of a high
permeability alloy slug between them
The two secondary coils are connected
in opposition in order to achieve a
wider region of linearity
The primary coil is sinusoidally excited,
with a frequency between 60 Hz and 20
kHz.
ad
e
LVDT
r.R
.B
.G
ho
ng
ad
e
Thealternatingmagneticfieldinducesnearlyequal
voltages and
inthesecondarycoils
Theoutputvoltage
=
Whentheslugissymmetricallyplaced,thetwosecondary
voltagesareequalandtheoutputsignaliszero
Linearvariabledifferentialtransformercharacteristics
includelinearity
overalargerange,achangeofphaseby180 whenthe
corepassesthroughthecenterposition,andsaturationon
theends
SpecificationsofcommerciallyavailableLVDTsinclude
sensitivitiesontheorderof0.5to2mVforadisplacement
of0.01mm/Vofprimaryvoltage,fullscaledisplacementof
0.1to250mm,andlinearityof0.25%
SensitivityforLVDTsismuchhigherthanthatforstrain
gauges
r.R
.B
.G
ho
ng
ad
e
(a)As movesthroughthe
nullposition,thephase
changes180,
whilethemagnitudeof is
proportionaltothe
magnitudeof
(b)Anordinaryrectifier
demodulatorcannot
distinguishbetween(a)and
(b),soaphasesensitive
demodulatorisrequired
r.R
.B
.G
ho
ng
Bloodflowthroughanexposedvesselcan
bemeasuredbymeansofan
electromagneticflowtransducer
Considerabloodvesselofdiameter filled
withbloodflowingwithauniformvelocity
Bloodvesselisplacedinauniform
magneticfield thatisperpendicularto
thedirectionofbloodflow
Negativelychargedanionandpositively
chargedcation particlesinthebloodwill
experienceaforce thatisnormalto
boththemagneticfieldandbloodflow
directions
ad
e
Electromagneticbloodflowtransducer
whereqistheelementarycharge(1.6x10
C)
ho
.G
ng
ad
e
Thesechargedparticleswillbedeflectedinopposite
directionsandwillmovealongthediameterofthe
bloodvesselsaccordingtothedirectionoftheforce
vector
Thismovementwillproduceanopposingforce
whichisequalto
r.R
.B
where isthenetelectricalfieldproducedbythe
displacementofthechargedparticlesand isthepotential
producedacrossthebloodvessel
Atequilibrium,thesetwoforceswillbeequal,hencethe
potentialdifference isgivenby
isproportionaltothevelocityofbloodthroughthevessel
r.R
.B
.G
ho
ng
ad
e
Electromagneticflowmeter
CapacitiveSensors
ThecapacitancebetweentwoparallelplatesofareaAseparatedby
distance is
ad
e
.(1)
ho
ng
r.R
.B
.G
ThesensitivityK ofacapacitivesensortochangesinplate
separation isfoundbydifferentiating(1)
Notethatthesensitivityincreasesastheplateseparationdecreases
ThepercentchangeinCaboutanyneutralpointisequaltothe
perunitchangein forsmalldisplacementsis
.G
ho
ng
ad
e
Capacitancesensorformeasuringdynamicdisplacement
changesandpressure
r.R
.B
Compliantplasticsofdifferentdielectricconstantsmay
beplacedbetweenfoillayerstoformacapacitivematto
beplacedonabed
Patientmovementgeneratescharge,whichisamplified
andfilteredtodisplayrespiratorymovementsfromthe
lungsandballistographic movementsfromtheheart
PiezoelectricSensors
r.R
.B
.G
ho
ng
ad
e
Piezoelectricsensorsareusedtomeasurephysiological
displacementsandrecordheartsounds
Piezoelectricmaterialsgenerateanelectricpotentialwhen
mechanicallystrained,andconverselyanelectricpotential
cancausephysicaldeformationofthematerial
Theprincipleofoperation:whenanasymmetricalcrystal
latticeisdistorted,achargereorientationtakesplace,
causingarelativedisplacementofnegativeandpositive
charges
Thedisplacedinternalchargesinducesurfacechargesof
oppositepolarityonoppositesidesofthecrystal
Surfacechargecanbedeterminedbymeasuringthe
differenceinvoltagebetweenelectrodesattachedtothe
surfaces
Where isthepiezoelectricconstant, /
ad
e
Assumeinfiniteleakageresistance,thetotalinducedcharge
isdirectlyproportionaltotheappliedforce
.G
ho
ng
Thechangeinvoltagecanbefoundbyassumingthatthe
systemactslikeaparallelplatecapacitorwherethevoltage
acrossthecapacitorischarge dividedbycapacitance
r.R
.B
ad
e
r.R
.B
ng
ho
Therearevariousmodesof
operationofpiezoelectricsensors,
dependingonthematerialandthe
crystallographicorientationofthe
plate
Thesemodesincludethethickness
orlongitudinalcompression,
transversalcompression,thickness
shearaction,andfaceshearaction
Alsoavailablearepiezoelectric
polymericfilms,suchas
polyvinylidene fluoride(PVDF)
Thesefilmsareverythin,
lightweightandpliant(easilyflexed
orbent),andtheycanbecuteasily
andadaptedtounevensurfaces
.G
ad
e
TemperatureMeasurements
ng
Temperatureisextremelyimportanttohuman
physiology
r.R
.B
.G
ho
example:lowtemperaturecanindicateonsetof
problems,e.g.,stroke
example:hightemperaturecanindicateinfection
Temperaturesensitiveenzymesandproteinscan
bedestroyedbyadversetemperatures
Temperaturemeasurementandregulationis
criticalinmanytreatmentplans
TemperatureSensorOptions
ThermoelectricDevices
ad
e
mostcommontypeiscalledThermocouple
canbemadesmallenoughtoplaceinsidecathetersorhypodermicneedles
ResistanceTemperatureDetectors(RTDs)
ho
ng
metalresistancechangeswithtemperature
Platinum,Nickel,Coppermetalsaretypicallyused
positivetemperaturecoefficients
.G
Thermistors(thermallysensitiveresistor)
r.R
.B
formedfromsemiconductormaterials,notmetals
oftencompositeofaceramicandametallicoxide(Mn,Co,CuorFe)
typicallyhavenegativetemperaturecoefficients
RadiantTemperatureSensors
photonenergychangeswithtemperature
measuredoptically(byphotodetector)
IntegratedCircuit(IC)TemperatureSensors
varioustemperatureeffectsinsiliconmanipulatedbycircuits
proportionaltoabsolutetemperature(PTAT)circuit:Sibandgap=
r.R
.B
.G
ho
ng
ad
e
PropertiesofTemperatureSensors
MetallicResistanceThermometers
r.R
.B
.G
ho
ng
ad
e
Theelectricresistanceofapieceofmetalorwiregenerally
increasesasthetemperatureofthatelectricconductor
increases
Alinearapproximationtothisrelationshipisgivenby
Itisimportanttomakesurethattheelectroniccircuitdoes
notpassalargecurrentthroughtheresistance
thermometertoprovideselfheatingduetotheJoule
conversionofelectricenergytoheat
r.R
.B
.G
ho
ng
ad
e
TemperatureCoefficientofResistancefor
CommonMetalsandAlloys
Thermocouples
Thermoelectricthermometryisbasedonthediscoveryof
Seebeck
ng
ad
e
dissimilarmetalsatdiff.temps.>signal
electromotiveforce(emf)isestablishedbythecontactoftwo
dissimilarmetalsatdifferenttemperatures
r.R
.B
.G
ho
Empiricalcalibrationdataareusuallycurvefittedwitha
powerseriesexpansionthatyieldstheSeebeck voltage
where isindegreesCelsiusand
thereferencejunctionismaintained
at0
Thermocouplefeatures:
ruggedandgoodforveryhightemperatures
notasaccurateasotherTempsensors(alsononlinearanddrift)
r.R
.B
.G
ho
ng
ad
e
ThermocoupleCircuits
(b)Thefirstlaw,homogeneouscircuits,statesthatinacircuit
composedofasinglehomogeneousmetal,onecannotmaintainan
electriccurrentbytheapplicationofheatalone
In(b)thenetemf atcdisthesameasin(a),regardlessofthefact
thatatemperaturedistribution(T3)existsalongoneofthewires
(A)
ad
e
ng
ho
r.R
.B
.G
Thesecondlaw,intermediatemetals,statesthatthe
netemf inacircuitconsistingofaninterconnectionof
anumberofunlikemetals,maintainedatthesame
temperature,iszero
Thepracticalimplicationofthisprincipleisthatlead
wiresmaybeattachedtothethermocouplewithout
affectingtheaccuracyofthemeasuredemf,provided
thatthenewlyformedjunctionsareatthesame
temperature
ad
e
ng
ho
.G
r.R
.B
Thethirdlaw,successiveorintermediatetemperatures,isillustratedin(d),
whereemf E1isgeneratedwhentwodissimilarmetalshavejunctionsat
temperaturesT1andT2andemf E2resultsfortemperaturesT2andT3.
Itfollowsthatanemf E1+E2resultsatcdwhenthejunctionsareat
temperaturesT1andT3
Thisprinciplemakesitpossibleforcalibrationcurvesderivedforagiven
referencejunctiontemperaturetobeusedtodeterminethecalibration
curvesforanotherreferencetemperature
Thethermoelectricsensitivity (alsocalledthe
thermoelectricpowerortheSeebeck coefficient)is
r.R
.B
.G
ho
ng
ad
e
UsingthermocouplewithcoldjunctioncompensatorLT1025
r.R
.B
.G
ho
ng
ad
e
CommonThermocouples
Thermistors
ad
e
Heavilyusedinbiomedicalapplications
r.R
.B
.G
ho
ng
baseresistivity:0.1to100ohmmeters
canbemadeverysmall,~500umdiameter
largesensitivitytotemperature(34%/C)
excellentlongtermstability
Resistancevs.temperature
keepcurrentlowtoavoidselfheating
ad
e
ng
ho
.G
r.R
.B
D
Thediagonallineswithapositiveslopegivelinearresistancevaluesandshowthe
degreeofthermistorlinearityatlowcurrents.
Theintersectionofthethermistorcurvesandthediagonallineswithnegativeslope
givethedevicepowerdissipation
SemiconductorThermometers
ThePTATVoltageandElectronicThermometry
r.R
.B
.G
ho
ng
ad
e
Thewelldefinedtemperaturedependenceofthe
diodevoltageisactuallyusedasthebasisformost
digitalthermometers
Wecanbuildasimpleelectronicthermometerin
whichtwoidenticaldiodesarebiasedbycurrent
sources and
Ifwecalculatethedifferencebetweenthediodevoltages
using
ho
ng
ad
e
wediscoveravoltagethatisdirectlyproportionaltoabsolute
temperature(PTAT),referredtoasthePTATvoltageVPTAT
r.R
.B
.G
ThePTATvoltagehasatemperaturecoefficientgivenby
Byusingtwodiodes,thetemperaturedependenceof has
beeneliminatedfromtheequation
Forexample,supposeT=295K,
=250A,and
=50
A,thenVPTAT=40.9mVwithatemperaturecoefficientof
+0.139mV/K.
ElectromagneticRadiationSpectrum
Visiblelightwavelength
~400700nm
Shorterwavelengths
ultraviolet,~100nm
xray,~1nm
gammarays,~0.1nm(=1)
Longerwavelengths
infraredIR:broadspectrum
nearIR,~1000nm=1m
thermalIR,~100m
farIR,~1mm
microwave,~1cm
radar,~110cm
TV&FMradio,~1m
AMradio,~100m
r.R
.B
.G
ho
ng
ad
e
RadiationThermometry
r.R
.B
.G
ho
ng
ad
e
Thereisaknownrelationshipbetweenthesurfacetemperatureof
anobjectanditsradiantpower
possibletomeasurethetemperatureofabodywithoutphysical
contactwithit
Atbodytemperatures,radiantspectruminfarinfrared
Medicalthermographyisatechniquewherebythetemperature
distributionofthebodyismappedwithasensitivityofafewtenths
ofakelvin.
Itisbasedontherecognitionthatskintemperaturecanvaryfrom
placetoplacedependingonthecellularorcirculatoryprocesses
occurringateachlocationinthebody.
Thermographyhasbeenusedfortheearlydetectionofbreast
cancer,butthemethodiscontroversial
Ithasalsobeenusedfordeterminingthelocationandextentof
arthriticdisturbances,forgaugingthedepthoftissuedestruction
fromfrostbiteandburns,andfordetectingvariousperipheral
circulatorydisorders(venousthrombosis,carotidarteryocclusions)
ad
e
ng
ho
.G
r.R
.B
D
HumanTemperatureMeasurement
Radiationthermometryisgoodfordetermining
internal(corebody)temperature
ad
e
ho
ng
measuresmagnitudeofinfraredradiationfromtympanic
membrane&surroundingearcanal
.G
tympanicmembraneisperfusedbythesamevasculatureasthe
hypothalamus,thebodysmainthermostat
doesnotneedtomakecontacttosettemperatureofthesensor
fastresponsetime,~0.1sec
accuracy~0.1C
independentofusertechniqueorpatientactivity
r.R
.B
advantagesoverthermometers,thermocouplesor
thermistors
requirescalibrationtargettomaintainaccuracy
ad
e
ng
ho
.G
r.R
.B
D
radiationthermometryisaninstrumentthatdeterminestheinternalorcorebody
temperatureofthehumanbymeasuringthemagnitudeofinfraredradiation
emittedfromthetympanicmembraneandsurroundingearcanal
FiberopticTemperatureSensor
ad
e
Sensoroperation
ho
ng
r.R
.B
.G
Canbemadesmallenoughforbiologicalimplantation
OpticalMeasurement
Widelyusedinmedicaldiagnosis
ad
e
clinicalchemistrylab:bloodandtissueanalysis
cardiaccatheterization:measureoxygensaturationofhemoglobin
source
filter
detector
ng
Opticalsystemcomponents
ho
Conventionalopticalsystem
Solidstate(semiconductor)opticalsystem
r.R
.B
.G
miniaturizeandsimplify
Optical/RadiationSources
Tungstenlamp
verycommonradiationsource
emissivityisfunctionofwavelength,
~40%for<1m(1000nm)
outputvariessignificantlywithtemperature
note2000Kand3000Kspectraonnextslide
highertemperatureshortenslifeoflampfilament
Arcdischargelamps
fluorescentlampsfilledwith,e.g.,carbon,mercury,sodium,xenon
morecompactw/highoutputperunitarea
Lightemittingdiodes(LED)
siliconbandgap~1.1eVnotveryefficientfordetection
GaAs,higherenergy(lowerwavelength),fast(~10ns)switching
GaP &GaAsP haveevenhigherenergy
LASER
commonlasers:HeNe,Argon(highpower,visualspectrum),CO2
semiconductorlasernotpreferred;energytoolow(infrared)
lasersalsousedtomendtears,e.g.,inretina
r.R
.B
.G
ho
ng
ad
e
Spectralcharacteristicsofsources,filters,detectors
3.
4.
2.
ACorning556glassfilterpassesa
bluewavelengthband
AKodak87gelatinfilterpasses
infraredandblocksvisible
wavelengths.
Germaniumlensespasslong
wavelengthsthatcannotbepassed
byglass
HemoglobinHb and
Oxyhemoglobin HbO passequally
at805nmandhavemaximal
differenceat660nm
1.
r.R
.B
(b) Filters
ng
3.
ho
2.
tungsten(W)at3000Khasabroad
spectraloutput.At2000K,outputis
loweratallwavelengthsandpeak
outputshiftstolongerwavelengths
Lightemittingdiodesyieldanarrow
spectraloutputwithGaAs inthe
infrared,GaP inthered,andGaAsP in
thegreen
Monochromaticoutputsfrom
commonlasersareshownbydashed
lines
.G
1.
ad
e
(a) Lightsources:
Spectralcharacteristicsofsources,filters,detectors
4.
5.
6.
d) Combination
ng
ho
3.
.G
2.
TheS4responseisatypical
phototuberesponse.
Theeyehasarelatively
narrowresponse,withcolors
indicatedbyVBGYOR.
CdS plusafilterhasa
responsethatclosely
matchesthatoftheeye.
Sipnjunctionsarewidely
used.
PbS isasensitiveinfrared
detector.
InSb isusefulinfarinfrared
r.R
.B
1.
ad
e
Detectors
c)
Indicatedcurvesfrom(a),(b),and
(c)aremultipliedateach
wavelengthtoyield(d),which
showshowwellsource,filter,
anddetectorarematched
focusenergyfromsourceintosmallerarea
placedtocollimateradiation(raysareparallel)
focusenergyfromtargetintodetector
ho
FiberOptics
r.R
.B
.G
efficienttransmissionofopticalsignalsover
distance
examplemedicalapplication:endoscope
Filters
ng
GeometricalOptics:Lenses
ad
e
OpticalTransmitter&Filters
controltransmittedpower
determinewavelengths(colors)transmitted
producewavelengthspectrum(diffraction
grating)
RadiationSensors
r.R
.B
.G
ho
ng
Spectralresponse
Si,noresponseabove1100nm
specialmaterials(InSb)
monitorskinradiation(300K)
Thermalsensors
transformsradiationintoheat
flatspectralresponsebutslow
subjecttoerrorfromchangesinambienttemperature
examplethermalsensors:thermistors,thermocouples
Quantumsensors
transformphotonenergyintoelectronrelease
sensitiveoveralimitedspectrumofwavelengths
examplequantumsensors:eye,photographicemulsion,sensorsbelow
Photoemissive sensors,e.g.phototube
Photoconductivecells
Photojunction sensors
Photovoltaicsensors
ad
e
Photoemissive Sensors
Construction&Operation
ng
ad
e
photocathodecoatedwithalkalimetal
incomingphotons(withenoughenergy,>1eVor1200nm)release
electronsfromphotocathode
releasedelectronsattractedtoanodeandformacurrentproportional
toincomingphotonenergy
ho
Example:phototube,liketheS4inthespectrumplots
Photomultiplier:phototubecombinedwithelectronamplifier
.G
very(themost?)sensitivephotodetector
r.R
.B
cooledtopreventthermalexcitationofelectrons
cancountindividualphotons
fastresponse,~10ns
comparetotheeye,whichcandetect~6photonswithin100ms
SolidStatePhotoelectricSensors
Photoconductivecells
Photoresistor
photosensitivecrystallinematerialsuchasCdS orPbS
incomingradiationcauseselectronstojumpbandgapand
produceelectronholepairs>lowerresistance
Photojunction sensors
incomingradiationgenerateselectronholepairsindiodedepletion
region
minimumdetectableenergybasedonbandgapofthediodesubstrate
(e.g.,Si)
canbeusedinphotovoltaicmodechange inopencircuitvoltageis
monitored
Photoncoupler
LEDphotodiodecombination
usedtoisolateelectricalcircuits
preventcurrentfromleakingoutofequipmentandintotheheart
ofapatient
r.R
.B
.G
ho
ng
ad
e
MEMSTransducers
MEMS=micro
electromechanical
system
ng
Microaccelerometer
ho
ad
e
miniature
transducerscreated
usingICfabrication
processes
Pressure
r.R
.B
gyroscope
Rotation
.G
cantileverbeam
suspendedmass
SensorCalibration
Sensorscanexhibitnonidealeffects
offset:nominaloutputnominalparametervalue
nonlinearity:outputnotlinearwithparameterchanges
crossparametersensitivity:secondaryoutputvariationwith,e.g.,
temperature
Calibration=adjustingoutputtomatchparameter
analogsignalconditioning
lookuptable
digitalcalibration
T=a+bV +cV2,T=temperature;V=sensorvoltage;
a,b,c =calibrationcoefficients
Compensation
removesecondarysensitivities
musthavesensitivitiescharacterized
canremovewithpolynomialevaluation
P=a+bV +cT +dVT +eV2,whereP=pressure,T=temperature
r.R
.B
.G
ho
ng
ad
e
ad
e
ng
ho
r.R
.B
.G
EndofLecture3
Lecture4a
ad
e
CardiovascularSystem
ng
I.Heartstructure&CardiacCycle
r.R
.B
.G
ho
II.Heartconductionsystem&ECG
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
TheCardiovascularSystem
ho
ng
ad
e
Heart:Oneofthemostimportantorganinthe
humanbody
Function:
r.R
.B
.G
Supplyoxygentoalltheparts(cells,tissues,
muscles,vitalorgans)ofthehumanbody
CollecttheexcretedCO2fromtheorgans
Describedmostlybycomparingitwithafluid
pump
Motivation
ad
e
Heartdisease
ng
Majorcauseofdeathsindevelopedanddeveloping
countries
r.R
.B
.G
ho
Useofengineeringmethodsandthedevelopmentof
instrumentationhavecontributedsubstantiallyto
progressmadeinrecentyearsinreducingdeathfrom
heartdiseases
Blood pressure , flow , and volume are measured by
using engineering techniques
Theelectrocardiogram,echocardiogramand
phonocardiogramaremeasuredandrecordedwith
electronicinstruments
TheHeartandthecardiovascular
system
r.R
.B
.G
ho
ng
ad
e
ng
ho
r.R
.B
.G
Theheartislocatedinthechest
betweenthelungsbehindthe
sternumandabovethe
diaphragm
Itissurroundedbythe
pericardium
Itssizeisaboutthatofafist,and
itsweightisabout250300g
Locatedabovetheheartarethe
greatvessels:
thesuperior
inferiorvenacava
thepulmonaryartery
thepulmonaryvein,a
theaorta
ad
e
LocationoftheHeart
Theaorticarchliesbehindtheheart
Theesophagusandthespineliefurtherbehindtheheart
ng
ho
.G
r.R
.B
ad
e
AnatomyoftheHeart
AnatomyoftheHeart
.G
ho
ng
ad
e
Thecardiacmusclefibersareorientedspirallyandaredividedintofourgroups
Twogroupsoffiberswindaroundtheoutsideofbothventricles
Beneaththesefibersathirdgroupwindsaroundbothventricles
Beneaththesefibersafourthgroupwindsonlyaroundtheleftventricle
Thefactthatcardiacmusclecellsareorientedmoretangentiallythanradially,
andthattheresistivityofthemuscleislowerinthedirectionofthefiberhas
importanceinelectrocardiography
Thehearthasfourvalves
r.R
.B
Tricuspid valve:betweentherightatriumandventriclelies
Mitralvalve:betweentheleftatriumandventricle
Pulmonary valve:betweentherightventricleandthepulmonaryartery,
Aorticvalve liesintheoutflowtractoftheleftventricle(controllingflowtotheaorta)
Thebloodreturnsfromthesystemiccirculationtotherightatriumandfrom
theregoesthroughthetricuspidvalvetotherightventricle
Itisejectedfromtherightventriclethroughthepulmonaryvalvetothelungs
Oxygenatedbloodreturnsfromthelungstotheleftatrium,andfromthere
throughthemitralvalvetotheleftventricle
Finallybloodispumpedthroughtheaorticvalvetotheaortaandthesystemic
circulation
r.R
.B
.G
ho
ng
ad
e
AnatomyoftheHeart
PathofBloodthroughtheHeart
r.R
.B
.G
ho
ng
ad
e
BloodthatislowinO2andhighinCO2enterstherightatriumthrough
thevenaecavae &coronarysinus
nextispumpedintothepulmonarycirculationafterbloodisoxygenated
inthelungs&someoftheCO2isremoved,itreturnstotheleftsideof
theheartthroughthepulmonaryveinsfromtheleftventricle
itmovesintotheaorta
gasexchangesoccurbetweenthebloodinthecapillariesandtheairin
thealveoliofthelungs
ORDERINWHICHBLOODFLOWS:
1. venaecavae &coronarysinus
2. rightatrium>tricuspidvalve
3. rightventricle>pulmonaryvalve>pulmonarytrunk
4. pulmonaryartery
5. pulmonaryvein
6. leftatrium>bicuspid(mitral)valve
7. leftventricle>aorticvalve
8. aorta
(lubbdupp)
r.R
.B
.G
ho
ng
ad
e
Bothpumpsaredividedintotwospacescalled
chamberssoyourheartisactuallya2barreled,4
chamberedpumper
Thetwosidesdonotworkindependently;theyare
preciselytimedasateamtomakethebestuseoftheir
pumpingpower(quiteefficient!)
As the heart pumps it makes a variety of clicks and
thumps; these are the sounds of the heart valves as
they click open & shut; each sound has a special
meaning
lubb isthesoundofthetricuspid&mitral(bicuspid)heartvalves
(onthetopchambers)shutting;
dupp isthesoundofthesemilunarheartvalvesclosing(these
heartvalvesshutoffthebigvesselsleavingtheheart)
Thehearthangsinthecenterofthechest
(mediastinum)
r.R
.B
.G
ho
ng
ad
e
Thecardiovascularsystem
ad
e
ng
ho
.G
r.R
.B
The
cardiovascular
system
Terminology
ad
e
Pulmonarycirculation
r.R
.B
Systemiccirculation
.G
ho
ng
Thecirculatorypathforbloodflowthroughthelungs
(functionofrightsideheart)
Pressuredifferencebetweenthearteriesandtheveinsis
small,lowresistance
Canbeconsideredasvolumepump
Thecirculatorysystemthatsuppliesoxygenandnutrients
tothecellsofthebody(functionofleftsideoftheheart)
Thissystemisahighresistancecircuitwithalargepressure
gradientbetweenthearteriesandveins
Canbeconsideredasapressurepump
r.R
.B
.G
ho
ng
ad
e
Themusclecontractionoftheleftheartislargerand
strongerthantherightheart,becauseofthegreater
pressuresrequiredforsystemiccirculation
Howeverthevolumeoftheblooddeliveredperunittime
bythetwosidesissamewhenmeasuredoverasufficiently
longintervaloftime
Theleftheartdevelopsapressureheadsufficienttocause
bloodflowtoallextremitiesofthebody
Thepumpingactionitselfisperformedbycontractionof
theheartmusclessurroundingeachchamberoftheheart
Thesemusclesreceivetheirownbloodsupplyfromthe
coronaryarteries ,whichsurroundtheheartlikeacrown(
corona)
Thecoronaryarterialsystemisaspecialbranchofthe
systemiccirculation
Pitfall!
ad
e
Whywecannotindiscriminatelyapproximatethesystem
withapumpandahydraulicsystem?
r.R
.B
.G
ho
ng
Thepipes,thearteriesandtheveinsarenotrigidbutflexible
Theyarecapableofhelpingandcontrollingbloodcirculationby
theirownmuscularactionandtheirownvalveandreceptor
system
BloodisnotapureNewtonianfluid;ratheritpossesses
propertiesthatdonotcomplywiththelawsgoverninghydraulic
motion
AlsothebloodrequireshelpfromthelungsforO2 andit
interactswiththelymphaticsystem
Manychemicalsandhormonesaffecttheoperationofthe
system
ad
e
ng
ho
r.R
.B
.G
Cardiovascular
circulation
Functioning
ad
e
Bloodenterstheheartontherightsidethrough
r.R
.B
.G
ho
ng
Superiorvenacava(comingfromupperbody
extremities)
Inferiorvenacava(comingfromlowerbody
extremities)
Incomingbloodfillsthestoragechamber,right
atrium
Coronarysinusalsoemptiesintorightatrium
(bloodthatcirculatesthroughtheheartitself)
r.R
.B
.G
ho
ng
ad
e
Whenrightatriumisfull,itcontractsand
forcesbloodthroughthetricuspidvalveinto
rightventricle
Rightventriclecontractstopumpbloodinto
pulmonarycirculationsystem
Tricuspidvalvecloseswhenpressurein
ventricleexceedsatrial pressure
Semilunarvalveopensandbloodisforced
intopulmonaryarteryandintothetwolungs
Inthealveoli oflungsredbloodcellsare
rechargedwithO2 andCO2 isexpelled
r.R
.B
.G
ho
ng
ad
e
Pulmonaryarterydividesmanytimesinto
smallerarteries(arterioles),whichsupply
bloodtoalveolarcapillaries,wherethe
exchangeofO2 andCO2 takesplace
Ontheothersideofthelungmassisasimilar
constructionwherecapillariesfeedintotiny
veins(venules),whichinturnformlargerveins
andultimatelyterminateintopulmonaryvein
Thispulmonaryveinreturnstheoxygenated
bloodtotheheart
r.R
.B
.G
ho
ng
ad
e
Thebloodenterstheleftatriumfrompulmonary
veinandispumpedthroughthemitral(bicuspid)
valveintoleftventriclebycontractionofatrial
muscles
Whenleftventricularmusclescontract,the
pressureproducedbycontractionmechanically
closesthemitralvalveandthebuildupof
pressureopenstheaorticvalve
Bloodisforcedintotheaorta whichsupplied
oxygenatedbloodtotheupperandthelower
extremitiesandinternalorgans
Thisactiontakesplacesynchronouslywithright
ventricle,wherethebloodisforcedintothe
pulmonaryartery
HeartspumpingCycle
ng
ho
.G
r.R
.B
Isdefinedasthe
periodof
contractionofthe
heartmuscles,
specificallythe
ventricular
muscles,atwhich
time,bloodis
pumpedintothe
pulmonaryartery
andtheaorta
ad
e
Systole
HeartspumpingCycle
ng
ho
.G
r.R
.B
Periodof
dilationof
theheart
cavitiesas
theyfillwill
blood
ad
e
Diastole
r.R
.B
.G
ho
ng
ad
e
Oncethebloodhasbeenpumpedintothe
arterialsystem,theheartrelaxes,pressurein
chambersdecreases,theoutletvalveclose
andinashorttimetheinletvalvesopenagain
torestartthediastoleandinitiatenewcyclein
theheart
Afterpassingthroughmanybifurcationsof
arteries,thebloodreachesvitalorgans,the
brainandtheextremities
Thelaststageofarterialsystemdividesinto
smallestarterioles
ThesearteriolesfeedintocapillarieswhereO2
issuppliedtocellsandCO2 isreceived
r.R
.B
.G
ho
ng
ad
e
Inturncapillariesjoinintovenules andthese
finallyforminferiorandsuperiorvenacava
Bloodsupplytotheheartitselfisfromaorta
throughcoronaryarteriesintoasimilar
capillarysystemtothecardiacveins
Thisbloodreturnstoheartchambersbythe
wayofcoronarysinus
Averageheartbeatrate
.G
ho
ng
75bpm
Mayvaryfrom60to85(sitting,standingposition)
Infantheartratemaybeashighas140bpm
HRincreaseswithheatexposure,physiologicaland
psychologicalfactors
r.R
.B
ad
e
Somefacts!
Heartpumpsabout5litersofbloodperminute
75to80%ofbloodvolumeinveins,20%in
arteries,remainingincapillaries
BloodSupplytotheHeart
r.R
.B
.G
ho
ng
ad
e
Heartmuscle(myocardium)needsblood
Coronaryarteriesbranchofffromsystemic
circulation&feedcapillariesthatpermeate
theheartmuscle(myocardium)
Whenblockageof toheartmusclesoccur
cardiacmusclesbegintodie&aheartattack
(myocardialinfarction)canoccurifblockageis
extensive
ad
e
ng
ho
.G
r.R
.B
TheConductionSystemoftheHeart
r.R
.B
.G
ho
ng
ad
e
Electricalstimulusneededtocauseheart
musclecontractions(systole)
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ho
ng
ad
e
ECGSummary
AlthoughrepresentationofanECGrecordingasascalartraceisillustratedin
Figure,severalothertechniquesforcardiacelectricalrepresentation,usually
closelylinkedtotherecordingtechnique,exist
ng
ho
.G
r.R
.B
GenesisofECG
waveformand
timingof
differentaction
potentialsfrom
differentregions
andspecialized
cellsoftheheart
andthe
corresponding
cardiaccycleof
theECGas
measuredonthe
bodysurface
manifestas
P,Q,R,S andT
points
ad
e
VariousComponentsoftheECGWaveform
TheApplicationAreasofECGDiagnosis
ho
ng
ad
e
6. Myocardialischemiaandinfarction
a. Ischemia
b. Infarction
7. Drugeffect
a. Digitalis
b. Quinidine
8. Electrolyteimbalance
a. Potassium
b. Calcium
9. Carditis
a. Pericarditis
b. Myocarditis
10. Pacemakermonitoring
r.R
.B
.G
1. Theelectricaxisoftheheart
2. Heartratemonitoring
3. Arrhythmias
a. Supraventriculararrhythmias
b. Ventriculararrhythmias
4. Disordersintheactivation
sequence
a. Atrioventricular conduction
defects(blocks)
b. Bundlebranchblock
c. WolffParkinsonWhite
syndrome
5. Increaseinwallthicknessorsize
oftheatriaandventricles
a. Atrialenlargement
(hypertrophy)
b. Ventricularenlargement
(hypertrophy)
r.R
.B
.G
ho
ng
ad
e
TheApplicationAreasofECGDiagnosis
ECGLeadSystems
ad
e
TheConventional12leadSystem
r.R
.B
.G
ho
ng
BipolarLimbLeads
WilsonCentralTerminal(WCT)
GoldbergerAugmentedLeads
PrecordialLeads
MasonandLikar LeadSystem(Modifiedleads)
TheCorrectedOrthogonalLeads(Franklead
system)
BipolarLimbLeads
Forcevectors:
r.R
.B
.G
ho
ng
ad
e
Themajorsequencesofdepolarizationoftheheartand
therelativevoltagesencounteredcanbeexplainedby
drawingaseriesofsummationvectors
Itisalsousefultodescribethedirectioninwhichthese
vectorsaretravelingbysuperimposingourdrawingona
360degreecompassrose
There are there important things that are the
underlying concepts of the lead systems:
1. The principle that impulses coming toward
an electrode produce positive deflections,
whereas impulses going away from an
electrode produce negative deflections.
2. The positions from which the various
electrodes look at the heart.
3. The sequence, direction, and relative
magnitude of the four major vectors of
cardiac depolarization and repolarization.
EinthovenlimbleadsandEinthoven
triangle
.G
ho
ng
ad
e
Intheyear1913,Einthovenetal.developedamethod
ofstudyingtheelectricalactivityoftheheartby
representingitgraphicallyinatwodimensional
geometricfigure,namely,anequilateraltriangle
Basedonseveraloversimplifyingassumptions
r.R
.B
Thebodyisahomogeneousvolumeconductor
Themeanofallelectricalforcescanbeconsideredas
originatinginanimaginarydipolelocatedintheelectrical
centeroftheheart
Electrodesplacedontherightarm(RA),leftarm(LA)and
leftfoot(LF)areusedtopickupthepotentialvariationson
theseextremitiestoformanequilateraltriangle
Einthoventriangle
ad
e
TheEinthovenlimbleads
(standardleads)are
definedinthefollowing
way
r.R
.B
where,
VI=voltageofleadI
VII=voltageofleadII
VIII=voltageofleadIII
L=potentialofleftarm
R=potentialofrightarm
F=potentialofleftfoot
.G
ho
ng
Thelimbleadsdescribethecardiac
electricalactivityinthreedifferent
directionsofthefrontalplane
ng
ho
.G
r.R
.B
FrankNormanWilson(18901952)investigated
howelectrocardiographicunipolarpotentials
couldbedefined
Measuredwithrespecttoaremotereference
(infinity)
Formedbyconnectinga5k resistorfromeach
terminalofthelimbleadstoacommonpoint
calledthecentralterminal
Wilsonsuggestedthatunipolarpotentials
shouldbemeasuredwithrespecttothis
terminalwhichapproximatesthepotentialat
infinity
TheWilsoncentralterminalistheaverageof
thelimbpotentials
Thetotalcurrentintothecentralterminalfrom
thelimbleadsmustaddtozerotosatisfythe
conservationofcurrent(KCL)
ad
e
WilsonCentralTerminal(WCT)
Hence,
r.R
.B
.G
ho
ng
ad
e
CircuitofWCTandthelocationimagespaceandthe
locationofWCT
GoldbergerAugmentedLeads
ad
e
Threeadditionallimbleads,VR,VL,andVF areobtainedbymeasuringthe
potentialbetweeneachlimbelectrodeandtheWilsoncentralterminal
Forinstance,themeasurementfromtheleftfootgives
ng
Goldbergerobservedthatthesesignalscanbeaugmentedbyomitting
thatresistancefromtheWilsoncentralterminal,whichisconnectedto
themeasurementelectrode
Threeleadsmaybereplacedwithanewsetofleadsthatarecalled
augmented leadsbecauseoftheaugmentationofthesignal
TheequationfortheaugmentedleadaVF is
r.R
.B
.G
ho
Augmentedsignaltobe50%largerthanthesignalwiththeWilsoncentral
terminalchosenasreference
GoldbergerAugmentedLeads
r.R
.B
.G
ho
ng
ad
e
Notethatthethreeaugmentedleads,aVR,aVL,andaVF,arefully
redundantwithrespecttothelimbleadsI,II,andIII
ad
e
ng
ho
.G
r.R
.B
ng
ho
.G
r.R
.B
Formeasuringthepotentialscloseto
theheart,Wilsonintroducedthe
precordialleads(chestleads)in1944
Theseleads,V1V6arelocatedover
theleftchest
ThepointsV1andV2arelocatedat
thefourthintercostalspaceonthe
rightandleftsideofthesternum
V4islocatedinthefifthintercostal
spaceatthemidclavicular line
V3islocatedbetweenthepointsV2
andV4
V5isatthesamehorizontallevelas
V4butontheanterioraxillaryline
V6isatthesamehorizontallevelas
V4butatthemidline
ad
e
PrecordialLeads
r.R
.B
.G
ho
ng
ad
e
Precordialchestleads areusedtorecordthevoltagedifference
betweentheseelectrodesandWilsonsCentralTerminus
r.R
.B
.G
ng
MasonandLikar recommendedmoving
thelimbelectrodesusedtorecordthe12
leadECGfromthelimbstothethoraxfor
exerciseelectrocardiography(1966)
The12leadsystemisusuallyusedforjust
longenoughtorecordafewheartcyclesor
beats(1015)seconds
Therecordedinformationisrepresentedas
12scalartracesdepictingthehearts
electricalactivityatthevarioussample
sites.
Interpretationofthe12leadECGisbased
uponexaminationoftheshapeandsize,or
amplitudeandduration,ofthevarious
componentsofeachscalartrace
Theincreasednumberofsamplesites,six
ofwhichareonthechestclosetothe
heart,allowsanexperttonotonly
determinethepresenceofdisease,but
alsothechambersorareasoftheheart
thatareaffected
ho
ad
e
MasonandLikar LeadSystem(Modifiedleads)
r.R
.B
.G
ng
ThisleadsystemisknownalsoasFrank
leadsystem
Sevenelectrodesplacedonthechest,back,
neckandleftfootareusedtoviewthe
heartfromtheleftside,frombelowand
fromthefront
Thiskindofleadsystemreflectsthe
electricalactivityinthethreeperpendicular
directionsX,Y,andZandtracesoutathree
dimensionalloopforeverycardiaccycleby
meansofthetimevariantcardiac
dominantvector
ThethreeprojectionsofthisloopontoXY,
XZandYZplanesarealsorecorded
Themorphologyoftheloops,their
directionofrotationandtheirareasarethe
mainspatialquantitiesthatimproveECG
baseddiagnosisofsomecardiac
pathologies,likemyocardialinfarction
Thisparticulartypeofrecordingisreferred
toasavectorcardiogram (VCG).
ho
ad
e
TheCorrectedOrthogonalLeads
ElectrocardiographBlockDiagram
ad
e
Lead
selector
Driven
rightleg
circuit
Isolation
circuit
Preamplifier
ng
Amplifier
protection
circuit
Baseline
restoration
ADC
Memory
Driver
amplifier
Recorder
printer
Isolated
power
supply
.G
ho
Auto
calibration
r.R
.B
Parallelcircuitsforsimultaneousrecordingsfromdifferentleads
Sensingelectrodes
Leadfaildetect
Amplifierprotection
circuit
Leadselector
Autocalibration
Preamplifier
Baselinerestoration
Drivenrightlegcircuit
Isolationcircuit
ADC&Memorysystem
Driveramplifier
Recorderprinter
Microcomputer
Controlsoftware
Leadfail
detect
Sensing
electrodes
Rightleg
electrode
Microcomputer
Operator
display
Control
program
Keyboard
ECGanalysis
program
FrequentProblems
.G
r.R
.B
ho
ng
Frequencydistortion
HighfrequencylossroundsthesharpedgesoftheQRScomplex.
Lowfrequencylosscandistortthebaseline(nolongerhorizontal)orcause
monophasic waveformstoappear biphasic.
Saturation/cutoffdistortion
Combinationofinputamplitude&offsetvoltagedrivesamplifierinto
saturation
Positivecase:clipsoffthetopoftheRwave
Negativecase:clipsofftheQ,S,PandTwaves
Groundloops
Patientsareconnectedtomultiplepiecesofequipment;eachhasaground
(powerlineorcommonroomgroundwire)
Ifmorethatoneinstrumenthasagroundelectrodeconnectedtothepatient,
agroundloopexists.Powerlinegroundcanbedifferentforeachitemof
equipment,sendingcurrentthroughthepatientandintroducingcommon
modenoise.
Openleadwires
Canbedetectedbyimpedancemonitoring.
ad
e
Unwantedvoltagetransients
Patientmovement
Electricalstimulation
signals,likedefibrillation
Amplifiersaturates
Firstorderrecoveryto
baseline
Recoverytimesetbylow
frequencycorner ofthe
bandpass amplifier
r.R
.B
.G
ho
ng
ad
e
Artifacts
Effectofavoltagetransientonan
ECGrecordedonan
electrocardiographinwhichthe
transientcausestheamplifierto
saturate,andafiniteperiodof
timeisrequiredforthechargeto
bleedoffenoughtobringtheECG
backintotheamplifiersactive
regionofoperation.Thisis
followedbyafirstorderrecovery
ofthesystem.
Upperfigure:couplingof50Hz/60Hzpowerlinenoise
Electricfieldcouplingbetweenpowergrid,instrument,patient,and
wiring.
Lowerfigure:couplingofelectromyographic (EMG)noise
ExampleoftensingchestmuscleswhileECGisbeingrecorded.
r.R
.B
.G
ho
ng
ad
e
Artifacts
PowerLineCoupling
Power line 230 V
ad
e
ng
Z1
ho
Z2
.G
C2
r.R
.B
Smallparasiticcapacitors connectthe
powerlinetotheRAandLAleads,
andthegroundedinstrumentcase
Smallacdisplacementcurrents Id1 and
Id2 aregenerated
Thebodyimpedanceisabout500
andcanbeneglected
vA vB =id1 Z1 id2 Z2(6.3)
IfId1 andId2 are approximatelyequal:
vA vB =id1 (Z1 Z2)(6.4)
=(6nA)(20K)
=120V
Remedies
Shieldelectrodes&connectto
electrocardiograph(grounding
tree)toreduceid
Reduceormatchtheelectrode
skinimpedances(minimizeZ1 Z2
)
ZG
Id1
Id2
C3
C1
A
B Electrocardiograph
G
Id1+ Id2
Amechanismofelectricfieldpickupofan
electrocardiographresultingfromthepower
line.Couplingcapacitancebetweenthehot
sideofthepowerlineandleadwirescauses
currenttoflowthroughskinelectrode
impedancesonitswaytoground.
PowerLineCoupling
Powerline
cm
Electrocardiograph
Z1
A
ng
Powerline iscoupledintothebody
Smallacdisplacementcurrent Idb is
generated,whichproducesacommon
modevoltage
vcm =idb ZG
(6.6)
=(0.2A)(50K)
=10mV
230V
cm
Zin
ho
ad
e
Cb
idb
Attheamplifierinputs:
(6.9)
vA vB =vcm (Z1 Z2)/Zin
=(10mV)(20K/5M
=40V
Remedies:
Reduceormatchtheelectrodeskin
impedances(minimizeZ1 Z2)
IncreaseZin
r.R
.B
.G
Z2
Zin
cm
G
ZG
idb
Currentflowsfromthepowerlinethroughthebody
andgroundimpedance,thuscreatingacommon
modevoltageeverywhereonthebody.Zin isnotonly
resistivebut,asaresultofRFbypasscapacitorsat
theamplifierinput,hasareactivecomponentas
well.
ad
e
MagneticFieldCoupling
r.R
.B
.G
ho
ng
Sources
Powerlines
Transformers
andballastsin
fluorescentlights
Remedies
Shielding
Routeleads
awayfrom
potentialsources
Reducethe
effectiveareaof
thesingleturn
coil(twistthe
leadwires)
Lecture4b
ng
ad
e
CardiovascularSystem
III.Phonocardiogram(PCG)
r.R
.B
.G
ho
IV.Electroencephalogram(EEG)
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
Introduction
r.R
.B
.G
ad
e
ng
Heartsoundsresultfromtheinterplayofthedynamiceventsassociated
withthecontractionandrelaxationoftheatriaandventricles,valve
movements,andbloodflow.
Canbeheardfromthechestthroughastethoscope,adevicecommonly
usedforscreeninganddiagnosisinprimaryhealthcare
Auscultation isthetermforlisteningtotheexternalsoundsofthebody,
usuallyusingastethoscope
Theartofevaluatingtheacousticpropertiesofheartsoundsand
murmurs,includingtheintensity,frequency,duration,number,andquality
ofthesounds,areknownascardiacauscultation.
Oneoftheoldestmeansforassessingtheheartcondition,especiallythe
functionofheartvalves
Thestethoscope(fromtheGreekwordstethos,meaning"chest"and
skopein,meaning"toexamine")inventedduringtheearlytwentieth
century,wasoneofthemostprimitivedevicesdesignedtoaidadoctorin
listeningtoheartsounds
Traditionalauscultationinvolvessubjectivejudgmentbytheclinicians,
whichintroducesvariabilityintheperceptionandinterpretationofthe
sounds,therebyaffectingdiagnosticaccuracy
ho
PHONOCARDIOGRAPHY TECHNIQUE
r.R
.B
.G
ho
ng
ad
e
Theauscultationoftheheartgivestheclinicianvaluable
informationaboutthefunctionalintegrityoftheheart
Additionaldetailscanbegatheredwhenthetemporal
relationshipsbetweentheheartsoundsandtheelectrical
andmechanicaleventsofthecardiaccyclearecompared
Thisapproachtotheanalysisofheartsoundsusinga
studyofthefrequencyspectraisknownas
phonocardiography
Thephonocardiogram isadevicecapableofobtaining
heartsoundsanddisplayingtheobtainedsignalsinthe
formofagraphdrawnwiththesignalamplitudeinone
axisandwithtimeintheother
r.R
.B
.G
ho
ng
ad
e
Blockdiagramofthegeneralbiomedicalsignalprocessingandanalysis,asanintegrative
approachforcomputeraideddiagnosissystem
CARDIOVASCULARPHYSIOLOGY(revisited)
ad
e
ng
ho
.G
r.R
.B
Theheartcanbeclassifiedfroma hemodynamicspointofviewasa
simplereciprocatingpump
Thepumpingchambershaveavariablevolumeandinputandoutput
ports
Aonewayvalveintheinputportisorientedsuchthatitopensonlywhen
thepressureintheinputchamberexceedsthepressurewithinthe
pumpingchamber
Anotheronewayvalveintheoutputportopensonlywhenpressurein
thepumpingchamberexceedsthepressureintheoutputchamber
Therodandcrankshaftwillcausethediaphragmtomovebackandforth
Thechambersvolumechangesasthepistonmoves,causingthepressure
withintoriseandfall
Intheheart,thechangeinvolumeistheresultofcontractionand
relaxationofthecardiacmusclethatmakesuptheventricularwalls
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ho
ng
ad
e
Themechanicalactivityoftheheartinvolvescontractionof
myocardialcells,opening/closingofvalves,andflowof
bloodtoandfromtheheartchambers
Thisactivityismodulatedbychangesinthecontractilityof
theheart,thecomplianceofthechamberwallsand
arteriesandthedevelopedpressuregradients
Themechanicalactivitycanbealsoexaminedusing
ultrasoundimaging
Theperipheralbloodflowsinthearteriesandveinsisalso
modulatedbymechanicalpropertiesofthetissue
TheflowofbloodcanbeimagedbyDopplerecho,andthe
pulsewavecanbecapturedinoneoftheperipheral
arteries
Thedifferenttypesofsignalsgiveusvariouspiecesof
informationaboutthecardiacactivity.Integratingthis
informationmayyieldabetterabilitytoassessthe
conditionofthecardiovascularsystem
CardiacSounds
r.R
.B
.G
ho
ng
ad
e
Audiblesoundsareproducedfromtheopeningandtheclosingof
theheartvalves,theflowofbloodintheheart,andthevibrationof
heartmuscles
Heartsoundsareshortlivedburstsofvibrationalenergyhavinga
transientcharacter
Theyareprimarilyassociatedwithvalvular and/orventricular
vibrations
Boththeirsiteoforiginandtheiroriginalintensitygovernsthe
radiationoftheheartsoundstothesurfaceofthechest
Therearefourseparatebasicsoundsthatoccurduringthe
sequenceofonecompletecardiaccycle
CardiacMurmurs
.G
r.R
.B
ho
ng
Murmursarevibrationscausedbyturbulenceinthebloodasitflows
throughsomenarroworificeortube.
Amurmurisoneofthemorecommonabnormalphenomenathatcanbe
detectedwithastethoscopeasomewhatprolonged'whoosh'thatcanbe
describedasblowing,rumbling,soft,harsh,andsoon
Murmursaresoundsrelatedtothenonlaminarflowofbloodintheheart
andthegreatvessels
Theyaredistinguishedfrombasicheartsoundsinthattheyarenoisyand
havealongerduration
Whileheartsoundshavealowfrequencyrangeandliemainlybelow200
Hz,murmursarecomposedofhigherfrequencycomponentsextendingup
to1000Hz
Mostheartmurmurscanreadilybeexplainedonthebasisofhighvelocity
floworabruptchangesinthecaliber(thediameteroftheinside)ofthe
vascularchannels
Typicalconditionsinthecardiovascularsystem,whichcausebloodflow
turbulence,arelocalobstructions,shunts,abruptchangesindiameter,
andvalveinsufficiency(valveisnotstrongenoughtopreventbackflow)
ad
e
S1
r.R
.B
.G
ho
ng
ad
e
Thefirstheartsound(S1)occursattheonsetofventricularsystole
Itcanbemostclearlyheardattheapexandthefourthintercostalspacesalong
theleftsternalborder
Itischaracterizedbyhigheramplitudeandlongerdurationincomparisonwith
otherheartsounds
Ithastwomajorhighfrequencycomponentsthatcanbeeasilyheardatbedside
AlthoughcontroversyexistsregardingthemechanismofS1,themostcompelling
evidenceindicatesthatthecomponentsresultfromtheclosureofthemitraland
tricuspidvalvesandthevibrationssetupinthevalvecusps,chordate,papillary,
muscles,andventricularwallsbeforeaorticejection
S1lastsforanaverageperiodof100200ms
Itsfrequencycomponentslieintherangeof10200Hz
TheacousticpropertiesofS1areabletorevealthestrengthofthemyocardial
systoleandthestatusoftheatrioventricular valvesfunction
Asaresultoftheasynchronousclosureofthetricuspidandmitralvalves,the
twocomponentsofS1areoftenseparatedbyatimedelayof2030ms
Thisdelayisknownasthe(split)inthemedicalcommunityandisofsignificant
diagnosticimportance
Anabnormallylargesplittingisoftenasignofheartproblem
S2
ad
e
ng
ho
.G
r.R
.B
Thesecondheartsound(S2)occurswithinashortperiodoncetheventriculardiastole
starts.
ItcoincideswiththecompletionoftheTwaveoftheelectrocardiogram(ECG)
S2consistsoftwohighfrequencycomponents,onebecauseoftheclosureoftheaortic
valveandtheotherbecauseoftheclosureofthepulmonaryvalve
Attheonsetofventriculardiastole,thesystolicejectionintotheaortaandthe
pulmonaryarterydeclinesandtherisingpressureinthesevesselsexceedsthepressure
intherespectiveventricles,thusreversingtheflowandcausingtheclosureoftheir
valves.
Thesecondheartsoundusuallyhashigherfrequencycomponentsascomparedwiththe
firstheartsound
Asaresultofthehigherpressureintheaortacomparedwiththepulmonaryartery,the
aorticvalvetendstoclosebeforethepulmonaryvalve,sothesecondheartsoundmay
haveanaudiblesplit
Innormalindividuals,respiratoryvariationsexistinthesplittingofS2
Duringexpirationphase,theintervalbetweenthetwocomponentsissmall(lessthan30
ms)
However,duringinspiration,thesplittingofthetwocomponentsisevident
Clinicalevaluationofthesecondheartsoundisabedsidetechniquethatisconsideredto
beamostvaluablescreeningtestforheartdisease
Manyheartdiseasesareassociatedwiththecharacteristicchangesintheintensitiesof
orthetimerelationbetweenthetwocomponentsofS2
S1andS2werebasicallythemaintwoheartsoundsthatwereusedformostofthe
clinicalassessmentbasedonthephonocardiographyauscultationprocedure
r.R
.B
.G
ho
ad
e
Thethirdandfourthheartsounds,alsocalledgallopsounds,arelow
frequencysoundsoccurringinearlyandlatediastole,respectively,under
highlyvariablephysiologicalandpathologicalconditions
Decelerationofmitralflowbyventricularwallsmayrepresentakey
mechanisminthegenesisofbothsounds
Thethirdheartsound(S3)occursintherapidfillingperiodofearly
diastole
Itisproducedbyvibrationsoftheventricularwallswhensuddenly
distendedbytherushofinflowresultingfromthepressuredifference
betweenventriclesandatria
TheaudibilityofS3maybephysiologicalinyoungpeopleorinsome
adults,butitispathologicalinpeoplewithcongestiveheartfailureor
ventriculardilatation
Thefourthheartsound(S4)occursinlatediastoleandjustbeforeS1
Itisproducedbyvibrationsinexpandingventricleswhenatriacontract.
Thus,S4israrelyheardinanormalheart
TheabnormallyaudibleS4resultsfromthereduceddistensibility(the
capabilityofbeingstretchedunderpressure)ofoneorbothventricles
Asaresultofthestiffventricles,theforceofatrialcontractionincreases,
causingsharpmovementoftheventricularwallandtheemissionofa
prominentS4
ng
S3&S4
ad
e
ng
ho
.G
r.R
.B
D
LUPP
DUBB
LUPP
MURMUR
DUBB
ad
e
ng
ho
.G
r.R
.B
Cardiaccycleeventsoccurringintheleftventricle
ho
ng
ad
e
Pressureprofileofthe
ventricleand
atrium
r.R
.B
D
Phonocardiographgy
signals
.G
Volumeprofileofthe
leftventricle
ad
e
ng
ho
.G
TheECG,PCG(lowandhigh
filtered),carotidpulse,
apexcardiogram,andlogicstates
(high=open)ofleftheartvalves,
mitralandaorticvalve,andright
heartvalves,tricuspidand
pulmonaryvalve
Leftheartmechanicalintervals
areindicatedbyverticallines:
isovolumic contraction(1),
ejection(2),isovolumic relaxation
(3),andfilling(4)(rapidfilling,
slowfilling,atrialcontraction)
ThelowfrequencyPCGshowsthe
fournormalheartsounds(I,II,III,
andIV)
InthehighfrequencytraceIIIand
IVhavedisappearedandsplitting
isvisibleinI[Ia andIb (andevena
smallIc duetoejection)]andinII
[IIA(aorticvalve)andIIP
(pulmonaryvalve)]
SystolicintervalsLVEP(oncarotid
curve)andQIIA(onECGandPCG)
areindicated
r.R
.B
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ho
ng
ad
e
Phonocardiographytracewith8successive
S1S2waveform.
r.R
.B
.G
ho
ng
ad
e
PCGsignalrecordingwithdifferentfilteringcoefficientfordifferent
correspondingheartsoundclass
PCGSIGNALSPECTRALANALYSIS
r.R
.B
.G
ho
ng
ad
e
Heartsoundsarecomplexandhighlynonstationarysignals
intheirnatureandhavebeenknowntobequasistationary
signalsforalongtime
Theheartbeatsassociatedwiththesesoundsarereacted
inthesignalbyperiodsofrelativelyhighactivityand
rhythmicenergystyle,alternatingwithcomparatively
intervalsoflowactivity
Accordingly,PCGSpectrometricpropertiescanbeextracted
bydifferentmethodsusing(e.g.,ShortTimeFourier
Transformation(STFT)),asitestimatesthepowerspectral
density(PSD)ofsuccessivewaveformandcomputedthese
transformationwillleadtoperiodicestimationofenergy
spikeswithintheacousticalwaveform
Classesofspectralanalysisused
Twobroadclassesofspectralanalysisapproaches
nonparametricmethods
parametric(modelbased)methods.
r.R
.B
.G
ho
ng
ad
e
Thenonparametricmethodssuchasperiodogram,Blackman
Tukey,andminimumvariancespectralestimatorsdonotimpose
anymodelassumptiononthedata,otherthanwidesense
stationarity
Theparametricspectralestimationapproaches,ontheotherhand,
assumethatthemeasurementdatasatisfyageneratingmodelby
whichthespectralestimationproblemisusuallyconvertedtothat
ofdeterminingtheparametersoftheassumedsignalmodel
Twokindsofmodelsarewidelyassumedandusedwithinthe
parametricmethods,accordingtodifferentspectralcharacteristics
ofthesignals:therationaltransferfunction(RTF)modelandthe
sinusoidalsignalmodel
TheRTFmodels,includingautocorrelation(AR),movingaverage
(MA),andautocorrelationmovingaverage(ARMA)typesare
usuallyusedtoanalyzethesignalswithcontinuousspectra,while
thesinusoidalsignalmodelisagoodapproximationofsignalswith
discretespectralpatterns
ad
e
ng
ho
.G
r.R
.B
ELECTROENCEPHALOGRAM(EEG)
The electroencephalogram (EEG) measures the activity of
large numbers (populations) of neurons.
r.R
.B
.G
ho
ng
ad
e
EEG
.G
ho
ng
ad
e
Spontaneousactivity ismeasuredonthescalporonthebrainandis
calledtheelectroencephalogram
TheamplitudeoftheEEGisabout100Vwhenmeasuredonthe
scalp,andabout12mVwhenmeasuredonthesurfaceofthebrain
Thebandwidthofthissignalisfromunder1Hztoabout50Hz
Asthephrase"spontaneousactivity"implies,thisactivitygoeson
continuouslyinthelivingindividual
r.R
.B
Evokedpotentials arethosecomponentsoftheEEGthatarisein
responsetoastimulus(whichmaybeelectric,auditory,visual,etc.)
Suchsignalsareusuallybelowthenoiselevelandthusnotreadily
distinguished,andonemustuseatrainofstimuliandsignal
averagingtoimprovethesignaltonoiseratio
Singleneuron behaviorcanbeexaminedthroughtheuseof
microelectrodeswhichimpalethecellsofinterest.Throughstudies
ofthesinglecell,onehopestobuildmodelsofcellnetworksthat
willreflectactualtissueproperties
r.R
.B
.G
ho
ng
ad
e
FrequencyspectrumofnormalEEG
EEGLEADSYSTEMS
Referencepointsare nasion,whichisthedelveatthetopofthenose,levelwiththeeyes;
inion,whichisthebonylumpatthebaseoftheskullonthemidlineatthebackofthehead
.G
ho
ng
Fromthesepoints,theskullperimetersaremeasuredinthetransverseandmedianplanes
Electrodelocationsaredeterminedbydividingtheseperimetersinto10%and20%intervals
Threeotherelectrodesareplacedoneachsideequidistantfromtheneighboringpoints
r.R
.B
ad
e
Inadditiontothe21electrodesoftheinternational1020system,
intermediate10%electrodepositionsarealsoused
Thelocationsandnomenclatureoftheseelectrodesarestandardizedby
theAmericanElectroencephalographicSociety
Inthisrecommendation,fourelectrodeshavedifferentnamescompared
tothe1020system;theseareT7,T8,P7,andP8.Theseelectrodesare
drawnblackwithwhitetextinthefigure
r.R
.B
.G
ho
ng
ad
e
r.R
.B
.G
ho
ng
ad
e
EEG
ad
e
ng
ho
.G
r.R
.B
EEG
ad
e
ng
ho
.G
r.R
.B
EEG
r.R
.B
.G
ho
ng
ad
e
EEG
ManyneuronsneedtosumtheiractivityinordertobedetectedbyEEGelectrodes.
Thetimingoftheiractivityiscrucial.Synchronizedneuralactivityproduceslarger
signals.
THEBEHAVIOROFTHEEEGSIGNAL
r.R
.B
.G
ho
ng
ad
e
Itispossibletodifferentiatealpha(),beta( ),delta(
),andtheta()wavesaswellasspikesassociatedwith
epilepsy
Thealphawaveshavethefrequencyspectrumof813
Hzandcanbemeasuredfromtheoccipitalregioninan
awakepersonwhentheeyesareclosed
Thefrequencybandofthebetawavesis1330Hz;
thesearedetectableovertheparietalandfrontallobes
Thedeltawaveshavethefrequencyrangeof0.54Hz
andaredetectableininfantsandsleepingadults
Thethetawaveshavethefrequencyrangeof48Hz
andareobtainedfromchildrenandsleepingadults
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ho
ng
ad
e
EEG
r.R
.B
.G
ho
ng
ad
e
EEG suffers from poor current source localization and the inverse
problem
EEG
r.R
.B
.G
ho
ng
ad
e
Power spectrum:
r.R
.B
.G
ho
ng
ad
e
SchematicofamplifierinputsforanalogEEGforalongitudinalbipolarmontage
Oneadditionalelectrodeinputthegroundisomittedforsimplicity
SinceanEEGvoltagesignalrepresentsadifferencebetweenthevoltagesattwo
electrodes,thedisplayoftheEEGforthereadingencephalographer maybesetupin
oneofseveralways
TherepresentationoftheEEGchannelsisreferredtoasamontage
AtypicaladulthumanEEGsignalisabout10Vto100Vinamplitudewhenmeasured
fromthescalp
r.R
.B
.G
ho
ng
ad
e
EEGOFA
NORMAL
HUMAN
ADULT
r.R
.B
.G
ho
ng
ad
e
EEGOFA
HUMAN
ADULT
SUFFERING
FROM
EPILEPSY
Lecture5
r.R
.B
.G
ho
ng
ad
e
I.XRayImaging
II.ComputedTomography
III.DiagnosticUltrasoundImaging
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
ad
e
ng
ho
.G
r.R
.B
I.XRayImaging
INTRODUCTIONTOXRAYIMAGING
r.R
.B
.G
ho
ng
ad
e
Xrayimagingisawellknownimagingmodalitythathas
beenusedforover100yearssinceRoentgendiscoveredX
raysbasedonhisobservationsoffluorescence
Xraysarehighenergyphotons
Theirgenerationcreatesincoherentbeamsthatexperience
insignificantscatterwhenpassingthroughvariousmedia
Asaresult,Xrayimagingisbasedonthroughtransmission
andanalysisoftheresultingXrayabsorptiondata
Xraysaredetectedthroughacombinationofaphosphor
screenandalightsensitivefilm
ad
e
ng
ho
.G
r.R
.B
ho
ng
ad
e
TypicalImagingChainfor
MedicalXraySystems
.G
processing
Collimator
r.R
.B
X-ray source
Object
Film
Image
r.R
.B
.G
ho
ng
ad
e
ElectromagneticRadiation
EMradiationcanbethoughtofasoscillatingelectricfield
whichgeneratesoscillatingmagneticfieldwhichgenerates
oscillatingelectricfieldandsoon.
Canalsobethoughtofasphotons(particles),asinCCD
detectionofvisiblelight.
ThisiscalledthewaveparticledualityofEMR.
Wavelength
= , wherev iscalledthephase
ng
ad
e
(lambda)iscalledwavelength,the
distancebetweentwoidenticalpoints
onawave
r.R
.B
.G
ho
speed(magnitudeofthephase
velocity)ofthewaveandf isthe
wave'sfrequency
InthecaseofEMradiation,the
equationbecomes= ,where c is
the speed of light: 3 x 108m/s
(nu)iscalledfrequency,thenumberofcyclesperunitoftime.
Itisinverselyproportionaltothewavelength.
Photons:review
r.R
.B
.G
ho
ng
ad
e
Photonsarelittlepacketsofenergy.
Eachphotonsenergyisproportionaltoits
frequency.
Aphotonsenergyisrepresentedbyh
E = h
r.R
.B
.G
ho
ng
ad
e
XRays
10-9 m (1 nm)
soft
XRayProduction
e-
r.R
.B
Anode (+)
.G
ho
e-
ng
ad
e
Cathode(-)
h
h
Electronsareacceleratedfromcathodetoanode.
Whenhighenergyelectronshitatomsofheavy
metals,theatomsproduceXrayphotons.
ad
e
ng
ho
.G
r.R
.B
XRayTube
GenerationofXrays
.G
r.R
.B
Bremsstrahlung
characteristicradiation
ho
ng
ad
e
Dependsonthermionicemissionandaccelerationofelectronsfrom
aheaterfilament
Duringthatprocess,electronsemittedfromcathodeare
acceleratedbyanodevoltage
KineticenergylossatananodeisconvertedtoXrays
Therelativepositionofanelectronwithrespecttothenucleus
determinesthefrequencyandenergyoftheemittedXray
XraysproducedinanXraytubecontaintwotypesofradiation
ThewordBremsstrahlungisretainedfromtheGermanlanguageto
describetheradiationthatisemittedwhenelectronsare
decelerated
ItischaracterizedbyacontinuousdistributionofXrayintensityand
shiftstowardhigherfrequencieswhentheenergyofthe
bombardingelectronsisincreased
CharacteristicXrays,ontheotherhand,producepeaksofintensity
atparticularphotonenergies
GeneratedXRayspectrum
r.R
.B
.G
ho
ng
ad
e
Inpractice,emittedradiationisfiltered,intentionallyor
not,producinghighpassfilterresponseaslowenergy
radiationis completelyattenuated
Asaresult,thefinalXrayspectrumhasbandpass
typecharacteristicswithseverallocalpeaks
superimposedonit
r.R
.B
.G
ho
ng
ad
e
Schematicrepresentationofa
standardXraysystem
Object
ho
.G
r.R
.B
ng
ad
e
ad
e
AttenuationCoefficient
ng
ho
Attenuation
Coefficient
.G
Bone
Muscle
Fat
0.1
r.R
.B
10
50
100
150
500
AttenuationCoefficient
ng
ad
e
Coefficientdependsonthepropertyofthe
material.
r.R
.B
.G
ho
Density (Bonehasahighdensitycomparedtosoft
tissues)
ChemicalMakeup (Leadblocksxrays;lead
screeningusedtoprotectpatient&technicians)
Exposure
(Capture)
Image
ho
ng
Processing
ad
e
Detector
r.R
.B
.G
Aspecialphotographicfilmisusedto
capturethexrayphotonswhichpassed
throughtheobject.
Thefilmisthenprocessed.
Filmturnsdarkwhereitwasexposedtox
rayphotons.
r.R
.B
.G
ho
ng
ad
e
TypicalXRayImages
Mammogram
Dental X-ray
ImageQualityFactors
ho
Object
ng
Energyofthephotons
Collimation
ad
e
Source
Detector
r.R
.B
.G
Attenuationcoefficient
Sourceobjectgeometry
Objectdetectorgeometry
Efficiency
AdvantagesofStandardDiagnostic
MedicalXrayImagingSystems
r.R
.B
.G
ho
ng
ad
e
Readilyavailable
Reasonablycheap
Simplesystemstomaintain
Manyexperiencedandtrainedpersonneldue
tothefactthattechnologyhasexistedfora
while
DisadvantagesofDiagnostic
MedicalXrayImagingSystems
r.R
.B
.G
ho
ng
ad
e
Exposuretoharmfulradiation.
Notmuchcontrastbetweendifferentsoft
tissues.
Imageisashadowgram(projectionimage)
withnodepthinformation.
ad
e
ng
r.R
.B
.G
ho
II.ComputedTomography
BasicConcepts
ad
e
highresolution
highcontrastimages
relativelysmallpatientexposure
permanentrecordoftheimage
Disadvantages
ng
ho
.G
Allxrayimagingsystemsconsistofanxraysource,acollimator,andanxray
detector
Diagnosticmedicalxraysystemsutilizeexternallygeneratedxrayswithenergies
of20150keV
Theshadowgraphimagesobtained aretheresultsofthevariationsinthe
intensityofthetransmittedxraybeamafterithaspassedthroughtissuesand
bodyfluidsofdifferentdensities
Advantages
r.R
.B
significantgeometricdistortion
inabilitytodiscerndepthinformation
incapabilityofprovidingrealtimeimagery
Conventionalradiography(XRayimaging)istheimagingmethodofchoicefor
suchtasksasdental,chest,andboneimagery
Whenthisprocedureisusedtoprojectthreedimensionalobjectsintoatwo
dimensionalplane,however,difficultiesareencountered
Structuresrepresentedonthefilmoverlap,anditbecomesdifficulttodistinguish
betweentissuesthataresimilarindensity.
Conventionalxraytechniquesareunabletoobtaindistinguishable/interpretable
imagesofthebrain,whichconsistsprimarilyofsofttissue
Xraytechniqueforvisualizingthreedimensional
structures
ad
e
ng
ho
.G
Knownasplanetomography
Theimagingofspecificplanesorcrosssectionswithinthebodybecame
possible
Thexraysourceismovedinonedirection,whilethephotographicfilm
(whichisplacedontheothersideofthebodyandpicksupthexrays)is
simultaneouslymovedintheotherdirection
Xraystravelcontinuously,changingpathsthroughthebody,eachray
passesthroughthesamepointontheplaneorcrosssectionofinterest
throughouttheexposure
Structuresinthedesiredplanearebroughtsharplyintofocusandare
displayedonfilm,whereasstructuresinalltheotherplanesareobscured
andshowuponlyasablur
Betterthanconventionalmethodsinrevealingthepositionanddetailsof
variousstructuresandinprovidingthreedimensionalinformationbysuch
atwodimensionalpresentation
r.R
.B
Tomography
Limitations
r.R
.B
.G
ho
ng
ad
e
doesnotreallylocalizea
singleplane,sincethereis
someerrorin
thedepthperception
obtained
largecontrastsinradio
densityareusually
requiredinordertoobtain
highqualityimagesthat
areeasytointerpret
Atomogramismadebyhavingthexray
xraydosesfor
tomographyarehigher
sourcemoveinonedirectionduringthe
thanroutineradiographs,
exposureandthefilmintheotherdirection
andbecausethe
Intheprojectedimage,onlyoneplanein
exposuresarelonger,
thebodyremainsstationarywithrespectto
patientmotionmay
themovingfilm
degradetheimage
Inthepicture,allotherplanesinthebody
content
areblurred
r.R
.B
ad
e
.G
ho
Consistsofascanninganddetection
system,acomputer,andadisplay
medium
Combinesimagereconstruction
techniqueswithxrayabsorption
measurementsinsuchawayasto
facilitatethedisplayofanyinternal
organintwodimensionalaxialslices
orbyreconstructionintheZaxisin
threedimensions
Acollimatedbeamofxraysis
directedthroughthesectionofbody
beingscannedtoadetectorthatis
locatedontheothersideofthe
patient
ng
ComputerizedTomography
Withanarrowlycollimatedsourceanddetectorsystem,itispossibletosenda
narrowbeamofxraystoaspecificdetectionsite
Someoftheenergyofthexraysisabsorbed,whiletheremaindercontinuesto
thedetectorandismeasured
Incomputerizedtomography,thedetectorsystemusuallyconsistsofacrystal
(suchascesiumiodideorcadmiumtungstate)thathastheabilitytoscintillate
oremitlightphotonswhenbombardedwithxrays
r.R
.B
.G
ho
ng
ad
e
Theintensityoftheselightphotonsorbundlesofenergyisinturn
measuredbyphotodetectorsandprovidesameasureoftheenergy
absorbed(ortransmitted)bythemediumthatispenetratedbythex
raybeam
Sincethexraysourceanddetectorsystemareusuallymountedona
frameorscanninggantry,theycanbemovedtogetheracrossand
aroundtheobjectbeingvisualized
Inearlydesigns,forexample,xrayabsorptionmeasurementswere
madeandrecordedateachrotationalpositiontraversedbythe
sourceanddetectorsystemcreatinganabsorptionprofileforthat
angularposition
Toobtainanotherabsorptionprofile,thescanninggantryholdingthe
xraysourceanddetectorwasthenrotatedthroughasmallangleand
anadditionalsetofabsorptionortransmissionmeasurementswas
recorded
Eachxrayprofileorprojectionobtainedinthisfashionisbasically
onedimensional
Itisaswideasthebodybutonlyasthickasthecrosssection
Example160x160picturematrix
r.R
.B
.G
ho
ng
ad
e
Exactnumberoftheseequallyspaced
positionsdeterminesthedimensionsto
berepresentedbythepictureelements
thatconstitutethedisplay
Absorptionmeasurementsfrom160
equallyspacedpositionsineach
translationarerequired
Eachonedimensionalarrayconstitutes
onexrayprofileorprojection
Toobtainthenextprofile,thescanning
unitisrotatedacertainnumberof
degreesaroundthepatient,and160
morelinearreadingsaretakenatthis
newposition
Eachofthemeasurementsobtainedbythe
Processisrepeatedagainandagain
precedingprocedureenterstheresident
untiltheunithasbeenrotatedafull
computerandisstoredinmemory
180
Oncealltheabsorptiondatahavebeen
Whenalltheprojectionshavebeen
obtainedandlocatedinthecomputers
collected,160x180,or28,800,
memory,thesoftwarepackagesdevelopedto
individualxrayintensitymeasurements
areavailabletoformareconstruction
analyzethedatabymeansofimage
ofacrosssectionofthepatientshead
reconstructionalgorithmsarecalledinto
orbody
action
Imagereconstruction
r.R
.B
.G
ho
ng
ad
e
Computerinitiallyestablishesagridconsistingofanumberofsmall
squaresforthecrosssectionofinterest,dependingonthesizeofthe
desireddisplay
Sincethecrosssectionofthebodyhasthickness,eachofthesesquares
representsavolumeoftissue,arectangularsolidwhoselengthis
determinedbytheslicethicknessandwhosewidthisdeterminedbythe
sizeofthematrix
Suchathreedimensionalblockoftissueisreferredtoasavoxel(or
volumeelement)andisonthedisplayintwodimensionsasapixel
(orpictureelement)
Duringthescanningprocess,eachvoxelisirradiatedbyanarrowbeam
ofxraysupto180times
Theabsorptioncausedbythatvoxelcontributestoupto180absorption
measurements,eachmeasurementpartofadifferentprojection
Eachvoxelaffectsauniquesetofabsorptionmeasurementstowhichit
hascontributed,thecomputercalculatesthetotalabsorptiondueto
thatvoxel
Usingthetotalabsorptionandthedimensionofthevoxel,theaverage
absorptioncoefficientofthetissuesinthatvoxelisdeterminedprecisely
anddisplayedinacorrespondingpixelasashadeofgrey
r.R
.B
.G
ho
ad
e
Thecrosssectionofinterestcanbeconsideredtobe
madeupofasetofblocksofmaterial
Eachblockhasanattenuatingeffectuponthe
passageofthexrayenergyorphotons,absorbing
someoftheincidentenergypassingthroughit
ThefirstblockabsorbsafractionA1 oftheincident
photons,thesecondafractionA2,andsoon,sothat
thenth blockabsorbsafractionAn
ThetotalfractionAabsorbedthroughalltheblocks
istheproductofallthefractions,whilethe
logarithmofthistotalabsorptionfractionisdefined
asthemeasuredabsorption
ng
OnlythemeasuredabsorptionfactorsA(1),A(2),A(3),
andA(4)wouldbeknown,butthiscanbesolvedsince
4simultaneousequationsand4unknowns
Toreconstructacrosssectioncontainingnrowsof
blocksandncolumns,itisnecessarytomakeat leastn
individualabsorptionmeasurementsfromatleastn
directions
.G
r.R
.B
ho
ng
Aparallelbeamofxraysisdirectedpastand
throughacylinderabsorbingsubstance
ashadowofthecylinderiscastonthexrayfilm
densityoftheexposedanddevelopedfilmalongthe
lineAAcanberegardedasaprojectionoftheobject
Ifaseriesofsuchradiographsaretakenatequally
spacedanglesaroundthecylinder,these
radiographsthenconstitutethesetofprojections
fromwhichthecrosssectionhastobe
reconstructed
Anapproximatereconstructioncanbereproduced
bydirectingparallelbeamsoflightthroughallthe
radiographsinturnfromthepositioninwhichthey
weretaken
Thecorrectcrosssectioncanbereconstructedby
backprojectingtheoriginalshadowandsubtracting
theresultofbackprojectingtwobeamsplacedon
eithersideoftheoriginalshadow
Mathematically,thisistheequivalentoftakingeach
transmissionvalueintheprojectionandsubtracting
fromitaquantityproportionaltoadjacentvalues
Thisprocessiscalledconvolution andisactuallyused
tomodifyprojections
ad
e
BackprojectionmethodofImageReconstruction
ad
e
ng
ho
.G
r.R
.B
Backprojection
r.R
.B
.G
ad
e
ng
Afile,usuallyknownasthepicturefile,
iscreatedinthecomputermemory
Containsanabsorptioncoefficientor
densityreadingforeachelementofthe
finalpicture
Resultantabsorptioncoefficientsfor
eachelementoftheimagecalculatedin
thismannercanthenbedisplayedas
graytonesorcolorscalesonavisual
display
Eachelementorpixelofthepicture
filehasavaluethatrepresentsthe
density(ormorepreciselytherelative
absorptioncoefficient)ofavolumein
thecrosssectionofthebodybeing
examined
ThescaledevelopedbyHounsfield
demonstratesthevaluesofabsorption
coefficientsthatrangefromair(1,000)
atthebottomofthescaletoboneat
thetop
ho
CTTechnology
ad
e
CTscannersareusuallyintegratedunitsconsistingof
threemajorelements:
r.R
.B
.G
ho
ng
1. Thescanninggantry,whichtakesthereadingsina
suitableformandquantityforapicturetobe
reconstructed
2. Thedatahandlingunit,whichconvertsthesereadings
intointelligiblepictureinformation,displaysthispicture
informationinavisualformat,andprovidesvarious
manipulativeaidstoenhancetheimageandthereby
assistthephysicianinformingadiagnosis
3. Astoragefacility,whichenablestheinformationtobe
examinedorreexaminedatanytimeaftertheactual
scan
TheScanningGantry
ad
e
ng
r.R
.B
Fourgenerationsof
scanninggantrydesigns.
Withmodernslipring
technology,third or
fourthgeneration
geometryallowsspiral
volumetricscanningusing
slicewidthsfrom1to10
mmandpixelmatrixesto
10,242
Typically,a50cmvolume
canbeimagedwitha
singlebreathhold
ho
Theobjectiveofthescanningsystemistoobtainenoughinformationto
reconstructanimageofthecrosssectionofinterest
CTscannershaveundergoneseveralmajorgantrydesignchanges
.G
ad
e
ng
ho
.G
r.R
.B
CTscanner
Mathematicalalgorithmsfortakingthe
attenuationprojectiondata
Canbeclassifiedintotwocategories:
Theiterativetechniques(alsoknownasthealgebraicreconstruction
technique[ART]),suchastheoneusedbyHounsfieldinthefirst
generationscanner,requireaninitialguessofthetwodimensional
patternofxrayabsorption
Theattenuationprojectiondatapredictedbythisguessarethen
calculatedandtheresultscomparedwiththemeasureddata
Thedifferencebetweenthemeasureddataandpredictedvaluesisusedin
aniterativemannersotheinitialguessismodifiedandthatdifference
goestozero
Ingeneral,alargenumberofiterationsarerequiredforconvergence,with
theprocessusuallyhaltedwhenthedifferencebetweenthecalculated
andthemeasureddataisbelowaspecifiederrorlimit
AnumberofdifferentversionsoftheARTweredevelopedandusedwith
first andsecondgenerationCATscanners
Latergenerationscannersusedanalyticreconstructiontechniques,since
theiterativemethodswerecomputationallyslowandhadconvergence
problemsinthepresenceofnoise
.G
r.R
.B
ho
ng
ad
e
iterative
Analytic
AnalyticAlgorithm
ng
AnalytictechniquesincludetheFouriertransform,backprojection,filteredback
projection,andconvolutionbackprojectionapproaches
Alloftheanalyticmethodsdifferfromtheiterativemethodsinthattheimageis
reconstructeddirectlyfromtheattenuationprojectiondata
Analytictechniquesusethecentralsectiontheoremandthetwodimensional
Fouriertransform
ad
e
.G
ho
Givenanimage
, , asingleprojectionistakenalongthe direction,forminga
projection
describedby
r.R
.B
Thisprojectionrepresentsanarrayoflineintegrals
ThetwodimensionalFouriertransformof
, is
givenby
IntheFourierdomain,alongtheline
transformbecomes
0,this
whichcanberewrittenas
ho
ng
ad
e
r.R
.B
.G
where 1 representsaonedimensionalFouriertransform
Itcanbeshownthatthetransformofeachprojectionformsaradiallinein
, , andtherefore
, canbedeterminedbytakingprojectionsatmany
anglesandtakingthesetransforms
When
, iscompletelydescribed,thereconstructedimagecanbefound
bytakingtheinverseFouriertransformtoobtain
,
CTscannersmaybecomparedwithoneanotherbyconsidering
thefollowingtenfactors:
r.R
.B
.G
ho
ng
ad
e
1. Gantrydesign,whichaffectsscanspeed,patientprocessingtime,andcosteffectiveness
2. Aperturesize,whichdeterminesthemaximumsizeofthepatientalongwiththeweight
carryingcapacityofthecouch
3. Thetypeofxraysource,whichaffectsthepatientradiationdoseandtheoveralllifeof
thescanningdevice
4. Xrayfanbeamangleandscanfield,whichaffectsresolution
5. Theslicethickness,aswellasthenumberofpulsesandtheangularrotationofthe
source,whichareimportantindeterminingresolution
6. Thenumberandtypesofdetectors,whicharecriticalparametersinimagequality
7. Thetypeofminicomputeremployed,whichisimportantinassessingsystemcapability
andflexibility
8. Thetypeofdatahandlingroutinesavailablewiththesystem,whichareimportantuser
andreliabilityconsiderations
9. Thestoragecapacityofthesystem,whichisimportantinascertainingtheaccessibilityof
thestoreddata
10. Upgradeabilityandconnectivitythatis,theyshouldbecapableofmodular
upgradeabilityandshouldcommunicatetoanyavailablenetwork
ad
e
ng
ho
r.R
.B
.G
III.DiagnosticUltrasoundImaging
Introduction
r.R
.B
.G
ho
ad
e
ng
ho
ng
ad
e
r.R
.B
.G
ho
ng
ad
e
Characteristics of Sound
Frequency
r.R
.B
.G
ng
ad
e
r.R
.B
.G
ho
r.R
.B
.G
ho
ng
ad
e
Characteristics of Sound
Speed
ad
e
ng
ho
.G
r.R
.B
Characteristics of Sound
Pressure, Intensity and the dB scale
r.R
.B
.G
ho
ad
e
ng
ad
e
ng
ho
.G
r.R
.B
.G
r.R
.B
ho
ng
ad
e
reflection
refraction
scattering
Absorption (attenuation)
Acoustic Impedance, Z
is equal to density of the material times speed of sound in the
material in which ultrasound travels, Z = c
= density (kg/m3) and c = speed of sound (m/sec)
measured in rayl (kg/m2sec)
Air and lung media have low values of Z, whereas bone and metal
have high values
Large differences in Z (air-filled lung and soft tissue) cause
reflection, small differences allow transmission of sound energy
The differences between acoustic impedance values at an interface
determines the amount of energy reflected at the interface
r.R
.B
.G
ho
ng
ad
e
Reflection
ad
e
ng
ho
.G
r.R
.B
Tissue reflections
ad
e
ng
ho
.G
r.R
.B
Refraction
r.R
.B
.G
ho
ng
ad
e
Scatter
ad
e
ng
ho
.G
r.R
.B
Acoustic scattering arises from objects within a tissue that are about
the size of the wavelength of the incident beam or smaller, and
represent a rough or nonspecular reflector surface
As frequency increases, the non-specular (diffuse scatter)
interactions increase, resulting in an increased attenuation and loss
of echo intensity
Scatter gives rise to the characteristic speckle patterns of various
organs, and is important in contributing to the gray-scale range in the
image
Attenuation
Ultrasound attenuation, the loss of energy with distance
travelled, is caused chiefly by scattering and tissue
absorption of the incident beam (dB)
The intensity loss per unit distance (dB/cm) is the
attenuation coefficient
Rule of thumb: attenuation in soft tissue is approx. 1
dB/cm/MHz
The attenuation coefficient is directly proportional to
and increases with frequency
Attenuation is medium dependent
r.R
.B
.G
ho
ng
ad
e
r.R
.B
ng
ho
.G
ad
e
Transducers
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
r.R
.B
.G
The thickness of a
piezoelectric crystal
determines the resonant
frequency of the
transducer
The operating resonant
frequency is determined
by the thickness of the
crystal equal to
wavelength (t=/2) of
emitted sound in the
crystal compound
Resonance transducers
transmit and receive
preferentially at a single
centre frequency
ho
ad
e
Damping Block
r.R
.B
.G
ho
ng
ng
ho
.G
r.R
.B
ad
e
Q factor
Matching Layer
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ho
ng
ad
e
ng
ho
64 to 128 elements
r.R
.B
.G
Simultaneous firing of
a small group of
approx. 20 elements
produces the
ultrasound beam
Rectangular field of
view produced for
linear and trapezoidal
for curvilinear array
transducers
ad
e
Transducer Arrays
ad
e
ng
ho
.G
r.R
.B
ng
ho
r.R
.B
.G
Is adjacent to the
transducer face and has a
converging beam profile
Convergence occurs
because of multiple
constructive and
destructive interference
patterns of the ultrasound
waves (pebble dropped in
a quiet pond)
ad
e
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
Angle of divergence
for non-focused
transducer is given by
sin() = 1.22/d
r.R
.B
.G
ho
ng
ad
e
ad
e
ng
ho
r.R
.B
.G
r.R
.B
.G
ho
ng
ad
e
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
For soft tissue (c = 1540 m/s or 0.154 cm/sec), the time delay
between the transmission pulse and the detection of the echo is
directly related to the depth of the interface as
c = 2D / time
Spatial Resolution
r.R
.B
.G
ho
ng
ad
e
ad
e
ng
ho
.G
r.R
.B
r.R
.B
.G
ng
ho
ad
e
ng
ho
.G
r.R
.B
ad
e
ad
e
ng
ho
.G
r.R
.B
ng
r.R
.B
.G
ho
ad
e
Display
ng
r.R
.B
.G
ho
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
Scan Converter
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
Lecture7
r.R
.B
.G
ho
DigitalsignalprocessingofBiosignals
Dr.R.B.Ghongade
DepartmentofE&TC,
V.I.I.T.,Pune411048
Objectives
r.R
.B
.G
ho
ng
ad
e
Noiseremoval
Clearlyunderstandthenature(analysis)
Diagnosisoftheunderlyingpathology
Aidintreatment(specificinstances)
SOURCESOFVARIABILITY:NOISE
r.R
.B
.G
ho
ng
ad
e
Noiseisaverygeneralandsomewhatrelative
term:noiseiswhatyoudonotwantand
signaliswhatyoudowant
Noiseisinherentinmostmeasurement
systemsandoftenthelimitingfactorinthe
performanceofamedicalinstrument
Manysignalprocessingtechniquesare
motivatedbythedesiretominimizethe
variability inthemeasurement
VARIABILITY
ho
ng
(1)physiologicalvariability
(2)environmentalnoiseorinterference
(3)transducerartifact
(4)electronicnoise
ad
e
Inbiomedicalmeasurements,variabilityhasfourdifferent
origins
r.R
.B
.G
Physiologicalvariabilityisduetothefactthattheinformation
youdesireisbasedonameasurementsubjecttobiological
influencesotherthanthoseofinterest
Forexample,assessmentofrespiratoryfunctionbasedon
themeasurementofbloodpO2couldbeconfoundedby
otherphysiologicalmechanismsthatalterbloodpO2
canbeaverydifficultproblemtosolve,sometimes
requiringatotallydifferentapproach
ad
e
Environmentalnoisecancomefromsources
externalorinternaltothebody
r.R
.B
.G
ho
ng
exampleisthemeasurementoffetalECGwherethe
desiredsignaliscorruptedbythemothersECG
notpossibletodescribethespecificcharacteristicsof
environmentalnoise,typicalnoisereduction
techniquessuchasfilteringarenotusuallysuccessful
canbereducedusingadaptivetechniques
techniquesdonotrequirepriorknowledgeofnoise
characteristics
ad
e
Transducerartifactisproducedwhenthe
transducerrespondstoenergymodalities
otherthanthatdesired
r.R
.B
.G
ho
ng
Forexample,recordingsofelectricalpotentials
usingelectrodesplacedontheskinaresensitive
tomotionartifact,wheretheelectrodesrespond
tomechanicalmovementaswellasthedesired
electricalsignal
Transducerartifactscansometimesbe
successfullyaddressedbymodificationsin
transducerdesign
r.R
.B
.G
ho
ng
ad
e
Electronicnoisehaswellknownsourcesand
characteristics
Electronicnoisefallsintotwobroadclasses
thermalorJohnsonnoise,
shotnoise
Johnsonnoiseisproducedprimarilyinresistoror
resistancematerials
Shotnoiseisrelatedtovoltagebarriersassociated
withsemiconductors
Bothsourcesproducenoisewithabroadrangeof
frequenciesoftenextendingfromDCto10121013 Hz
Suchabroadspectrumnoiseisreferredtoaswhite
noisesinceitcontainsenergyatallfrequencies
ad
e
ng
ho
.G
r.R
.B
JohnsonNoise
ng
ad
e
Johnsonorthermalnoiseisproducedbyresistance
sources,andtheamountofnoisegeneratedisrelatedto
theresistanceandtothetemperature:
r.R
.B
.G
ho
whereRistheresistanceinohms,TthetemperatureindegreesKelvin,andk
isBoltzmans constant(k=1.38 1023J/K).*Bisthebandwidth,orrangeof
frequencies,thatisallowedtopassthroughthemeasurementsystem
(Thesystembandwidthisdeterminedbythefiltercharacteristicsinthesystem,usually
theanalogfilteringinthesystem)
Ifnoisecurrentisofinterest,theequationforJohnsonnoise
currentcanbeobtainedas:
r.R
.B
.G
ho
ng
ad
e
SinceJohnsonnoiseisspreadevenlyoverall
frequencies,itisnotpossibletocalculatea
noisevoltageorcurrentwithoutspecifyingB,
thefrequencyrange
Sincethebandwidthisnotalwaysknownin
advance,itiscommontodescribearelative
noise;specifically,thenoisethatwouldoccur
ifthebandwidthwere1.0Hz
Suchrelativenoisespecificationcanbe
identifiedbytheunusualunitsrequired:
or
Shotnoise
ho
ng
ad
e
Shotnoiseisdefinedasacurrentnoiseandis
proportionaltothebaselinecurrentthrougha
semiconductorjunction
r.R
.B
.G
Noiseisspreadacrossallfrequencies,the
bandwidth,BW,mustbespecifiedtoobtainaspecific
value,orarelativenoisecanbespecifiedin
MultipleNoiseSources
r.R
.B
.G
ho
ng
ad
e
Whenmultiplenoisesourcesarepresent,asis
oftenthecase,theirvoltageorcurrent
contributionstothetotalnoiseaddasthe
squarerootofthesumofthesquares,
assumingthattheindividualnoisesourcesare
independent
SignaltoNoiseRatio
r.R
.B
.G
ho
ng
ad
e
Mostwaveformsconsistofsignalplusnoisemixedtogether
Signalandnoisearerelativeterms,relativetothetaskat
hand:thesignalisthatportionofthewaveformofinterest
whilethenoiseiseverythingelse
Goalofsignalprocessingistoseparateoutsignalfromnoise,
toidentifythepresenceofasignalburiedinnoise,orto
detectfeaturesofasignalburiedinnoise
Therelativeamountofsignalandnoisepresentinawaveform
isusuallyquantifiedbythesignaltonoiseratio,SNR
Istheratioofsignaltonoise,bothmeasuredinRMS(root
meansquared)amplitude
r.R
.B
.G
ho
ng
ad
e
Itisdifficulttodetect
presenceofthesignal
visuallywhentheSNRis
3db,andimpossible
whentheSNRis10db
TheabilitytodetectsignalswithlowSNRisthegoalandmotivationformanyofthe
signalprocessingtools
ANALOGTODIGITALCONVERSION:BASIC
CONCEPTS
.G
r.R
.B
slicingintime
slicinginamplitude
ho
ng
ad
e
Convertsananalogvoltagetoanequivalentdigitalnumber
Analogorcontinuouswaveform,x(t),isconvertedintoa
discretewaveform,x(n),afunctionofrealnumbersthatare
definedonlyatdiscreteintegers,n
Requires
Slicingthesignalintodiscretepointsintimeistermedtime
samplingorsimplysampling
Timeslicingsamplesthecontinuouswaveform,x(t),at
discretepointsintime,nTs,whereTsisthesampleinterval
SincethebinaryoutputoftheADCisadiscreteintegerwhile
theanalogsignalhasacontinuousrangeofvalues,analogto
digitalconversionalsorequirestheanalogsignaltobesliced
intodiscretelevels,aprocesstermedquantization
ad
e
ng
ho
.G
r.R
.B
Assumingthattheerrorisuniformlydistributedbetween/2+/2,the
variance,,is:
r.R
.B
.G
ng
Thenumberofbitsusedfor
conversionsetsalowerlimitonthe
resolution,andalsodeterminesthe
quantizationerror
Thiserrorcanbethoughtofasa
noiseprocessaddedtothesignal
Ifasufficientnumberofquantization
levelsexist(sayN>64),the
distortionproducedbyquantization
errormaybemodeledasadditive
independentwhitenoisewithzero
meanwiththevariancedetermined
bythequantizationstepsize,=
VMAX/2N
ho
ad
e
QuantizationError
Assumingauniformdistribution,theRMSvalueofthenoisewouldbejusttwicethe
standarddeviation,
NoiseRepresentation
ng
ho
probabilitydistribution
rangeofvariability
frequencycharacteristics
ad
e
Noiseisusuallyrepresentedasarandomvariable,x(n)
Describingnoiseasafunctionoftimeisnotveryuseful
Morecommontodiscussotherpropertiesofnoisesuch:
r.R
.B
.G
Noisecantakeonavarietyofdifferentprobability
distributions
CentralLimitTheoremimpliesthatmostnoisewillhavea
Gaussianornormaldistribution
TheCentralLimitTheoremstatesthatwhennoiseis
generatedbyalargenumberofindependentsourcesitwill
haveaGaussianprobabilitydistributionregardlessofthe
probabilitydistributioncharacteristicsoftheindividual
sources
ad
e
ng
ho
.G
r.R
.B
(A)Thedistributionof
20,000uniformly
distributedrandom
numbers.
(B)Thedistributionof
20,000numbers,eachof
whichistheaverageof
twouniformlydistributed
randomnumbers
(C)and(D)The
distributionobtained
when3and8random
numbers,stilluniformly
distributed,areaveraged
together
Althoughtheunderlyingdistributionisuniform,theaveragesofthese
uniformlydistributednumberstendtowardaGaussiandistribution
TheprobabilityofaGaussiandistributedvariable,x,isspecifiedin
thewellknownnormalorGaussiandistributionequation
ad
e
Twoimportantpropertiesofarandomvariableareitsmean,or
averagevalue,anditsvariance,theterm2
.G
ho
ng
ThemeanvalueofadiscretearrayofNsamplesisevaluatedas:
r.R
.B
Thesamplevariance,2,iscalculatedas
andthestandarddeviation,,isjustthesquarerootofthevariance
Normalizingthestandarddeviationorvarianceby1/(N1)produces
thebestestimateofthevariance,ifxisasamplefromaGaussian
distribution
.G
Butthestandarddeviationisthesquarerootofthevarianceandthestandard
deviationsaddasthe
timestheaveragestandarddeviation
Accordingly,themeanstandarddeviationistheaverageoftheindividual
standarddeviationsdividedby
r.R
.B
ho
ng
Whenmultiplemeasurementsaremade,multiplerandomvariablescanbe
generated
Ifthesevariablesarecombinedoraddedtogether,themeansaddsothatthe
resultantrandomvariableissimplythemean,oraverage,oftheindividual
means
Thesameistrueforthevariancethevariancesaddandtheaveragevariance
isthemeanoftheindividualvariances:
ad
e
Thusaveragingnoisefromdifferentsensors,ormultipleobservationsfromthe
samesource,willreducethestandarddeviationofthenoisebythesquareroot
ofthenumberofaverages
SpectralcharacteristicsofNoise
r.R
.B
.G
ho
ng
ad
e
Energydistributionmayvarywithfrequency
Frequencycharacteristicsofthenoisearerelatedto
howwell,oneinstantaneousvalueofnoisecorrelates
withtheadjacentinstantaneousvalues:fordigitized
datahowmuchonedatapointiscorrelatedwithits
neighbors
Ifthenoisehassomuchrandomnessthateachpointis
independentofitsneighbors,thenithasaflatspectral
characteristicandviceversa,calledaswhitenoise
Whenwhitenoiseisfiltered,itbecomesbandlimited
andisreferredtoascolorednoise
ENSEMBLEAVERAGING
r.R
.B
.G
ho
ng
ad
e
Averagingcanbeasimple,yetpowerfulsignal
processingtechniqueforreducingnoisewhenmultiple
observationsofthesignalarepossible
Inmanybiomedicalapplications,themultiple
observationscomefromrepeatedresponsestothe
samestimulus
Inensembleaveraging,agroup,orensemble,oftime
responsesareaveragedtogetheronapointbypoint
basis;thatis,anaveragesignalisconstructedbytaking
theaverage,foreachpointintime,overallsignalsin
theensemble
Twoessentialrequirements
theabilitytoobtainmultipleobservations
referencesignalcloselytimelinkedtotheresponse
r.R
.B
.G
ho
ng
ad
e
Anensembleofindividual(vergence)eyemovement
responsestoastepchangeinstimulus
DATAFUNCTIONSANDTRANSFORMS
ad
e
Insignalprocessing,mostfunctionsfallintotwo
categories
ng
waveforms,images,orotherdata
entitiesthatoperateonwaveforms,images,orotherdata
r.R
.B
.G
ho
canbefurtherdividedintofunctionsthatmodifythedata,and
functionsusedtoanalyzeorprobethedata
Example1:filtercoefficients(modifythespectralcontentofa
waveform)
Example2: FourierTransformusesfunctions(harmonicallyrelated
sinusoids)toanalyzethespectralcontentofawaveform
Functionsthatmodifydataarealsotermedoperations
ortransformations
Atransformcanbethoughtofasaremappingofthe
originaldataintoafunctionthatprovidesmore
informationthantheoriginal
Howtransformswork!
Transformsareachievedbycomparingthesignalofinterestwithsome
sortofprobingfunction
Comparisontakestheformofacorrelation(producedbymultiplication)
thatisaveraged(orintegrated)overthedurationofthewaveform,or
someportionofthewaveform
ho
ng
ad
e
Afamilyofprobingfunctionsisalsotermedabasis
Fordiscretefunctions,aprobingfunctionconsistsofasequenceofvalues,
orvector,andtheintegralbecomessummationoverafiniterange
r.R
.B
.G
wherex(n) isthediscretewaveformandfm(n)isadiscreteversionofthefamily
ofprobingfunctions.
Thisequationassumestheprobeandwaveformfunctionsarethesamelength
r.R
.B
.G
ho
ng
ad
e
Wheneitherx(t)orfm(t) areofinfinitelength,theymustbetruncated
Alsoifthelengthoftheprobingfunction,fm(n),isshorterthanthe
waveform,x(n),thenx(n) mustbeshortenedinsomeway
Canbeshortenedbysimpletruncationorbymultiplyingthefunctionby
anotherfunctionthathaszerovaluebeyondthedesiredlength
Afunctionusedtoshortenanotherfunctionistermedawindowfunction
Consequencesofthisartificialshorteningwilldependonthespecific
windowfunctionused
whereW(n)isthewindowfunction
FiniteSupport
r.R
.B
.G
ho
ng
Iftheprobingfunctionisoffinitelength(finitesupport)andthislengthis
shorterthanthewaveform,thenitmightbeappropriatetotranslateor
slideitoverthesignalandperformthecomparison(correlation,or
multiplication)atvariousrelativepositionsbetweenthewaveformand
probingfunction
Theoutputwouldbeafamilyoffunctions,oratwovariablefunction,
whereonevariablecorrespondstotherelativepositionbetweenthetwo
functionsandtheothertothespecificfamilymember
ad
e
wherethevariablek indicatestherelativepositionbetweenthetwofunctions
andm isthefamilymemberasintheaboveequations
Approachcanbeusedforlongoreveninfiniteprobingfunctions,
providedtheprobingfunctionitselfisshortenedbywindowingtoalength
thatislessthanthewaveform
Theshortenedprobingfunctioncanbetranslatedacrossthewaveformin
thesamemannerasaprobingfunctionthatisnaturallyshort
UsedinSTFT
ad
e
ng
ho
.G
r.R
.B
Projections
ad
e
ng
ho
.G
r.R
.B
Whentheprobingfunctionconsistsofafamilyoffunctions,thenthescalar
productoperationscanbethoughtofasprojectingthewaveformvector
ontovectorsrepresentingthevariousfamilymembers
Inthisvectorbasedconceptualization,theprobingfunctionfamily,orbasis,
canbethoughtofastheaxesofacoordinatesystem
Hencethemotivationbehinddevelopmentofprobingfunctionsthathave
familymembersthatareorthogonalororthonormal sothatthescalar
productcomputations(orprojections)canbedoneoneachaxes(i.e.,on
eachfamilymember)independentlyoftheothers
Allofthediscreteequations(discussedsofar)haveonethingincommon:
theyallfeaturethemultiplicationoftwo(orsometimesthree)functions
andthesummationoftheproductoversomefiniteinterval
Thismultiplicationandsummationisthesameasscalarproductofthe
twovectors
ad
e
ng
ho
.G
r.R
.B
CONVOLUTION
Importantconceptinlinearsystemstheory,solvingtheneedforatime
domainoperationequivalenttotheTransferFunction
Convolutioncanbeusedtodefineageneralinputoutputrelationshipin
thetimedomainanalogoustotheTransferFunctioninthefrequency
domain
r.R
.B
.G
ho
ng
ad
e
Theinput,x(t),theoutput,y(t),andthefunctionlinkingthetwothrough
convolution,h(t),areallfunctionsoftime;hence, convolutionisatime
domainoperation
Basicconceptbehindconvolutionissuperposition
Thefirststepistodetermineatimefunction,h(t),thattellshowthe
systemrespondstoaninfinitelyshortsegmentoftheinputwaveform
Ifsuperpositionholds,thentheoutputcanbedeterminedbysumming
(integrating)alltheresponsecontributionscalculatedfromtheshort
segments
ImpulseResponse
r.R
.B
.G
ho
ng
ad
e
Thewayinwhichalinearsystemrespondstoaninfinitelyshort
segmentofdatacanbedeterminedsimplybynotingthesystems
responsetoaninfinitelyshortinput,aninfinitelyshortpulse
Aninfinitelyshortpulse(oronethatisatleastshortcomparedto
thedynamicsofthesystem)istermedanimpulseordeltafunction
(commonlydenoted(t)),andtheresponseitproducesistermed
theimpulseresponse,h(t).
Giventhattheimpulseresponsedescribestheresponseofthe
systemtoaninfinitelyshortsegmentofdata,andanyinputcanbe
viewedasaninfinite
stringofsuchinfinitesimalsegments,theimpulseresponsecanbe
usedtodeterminetheoutputofthesystemtoanyinput
Responseproducedbyaninfinitelysmalldatasegmentissimply
thisimpulseresponsescaledbythemagnitudeofthatdata
segment
Thecontributionofeachinfinitelysmallsegmentcanbesummed,
orintegrated,tofindtheresponsecreatedbyallthesegments
ad
e
ng
ho
.G
r.R
.B
MathematicalDescriptionofConvolution
Statedmathematically,theoutputy(t),toanyinput,x(t)isgivenby:
ng
ad
e
Todeterminetheimpulseofeachinfinitelysmalldatasegment,the
impulseresponseisshiftedatimewithrespecttotheinput,thenscaled
(i.e.,multiplied)bythemagnitudeoftheinputatthatpointintime
Itdoesnotmatterwhichfunction,theinputortheimpulseresponse,is
shifted
Shiftingandmultiplicationissometimesreferredtoasthelagproduct
Fordiscretesignals,theintegrationbecomesasummationandthe
convolutionequationbecomes:
r.R
.B
.G
ho
Correlation
r.R
.B
.G
ho
ng
ad
e
Wordcorrelationconveyssimilarity:howonethingislikeanother
Mathematically,correlationsareobtainedbymultiplying andnormalizing
Covarianceandcorrelationusemultiplicationtocomparethelinear
relationshipbetweentwovariables,butincorrelationthecoefficientsare
normalizedtofallbetweenzeroandone
Becauseofnormalizationcorrelationcoefficients areinsensitive to
variationsinthegainofthedataacquisitionprocessorthescalingofthe
variables
Canbeappliedto
twoormorewaveforms
multipleobservationsofthesamesource
multiplesegmentsofthesamewaveform
Thecorrelationfunctionisthelaggedproductoftwowaveforms:
Alsocalledcross
correlation
Autocorrelation
r.R
.B
.G
ho
ng
ad
e
Specialcaseofthecorrelationfunctionoccurswhenthe
comparisonisbetweentwowaveformsthatareoneinthesame;
thatis,afunctioniscorrelatedwithdifferentshiftsofitself
Providesadescriptionofhowsimilara waveformistoitselfat
varioustimeshifts,ortimelags
Autocorrelationfunctionwillnaturallybemaximumforzerolag(n=
0)becauseatzerolagthecomparisonisbetweenidentical
waveforms
Usuallytheautocorrelationisscaledsothatthecorrelationatzero
lagis1
Functionmustbesymmetricaboutn=0,sinceshiftingoneversion
ofthesamewaveforminthenegativedirectionisthesameas
shiftingtheotherversioninthepositivedirection
Relatedtothebandwidthofthewaveform
Thesharperthepeakoftheautocorrelationfunctionthebroader
thebandwidth
ad
e
ng
ho
.G
r.R
.B
Crosscovariance
r.R
.B
.G
ho
ng
ad
e
Sameascrosscorrelationfunctionexceptthatthemeans
havebeenremovedfromthedatabeforecalculation
Thetermscorrelationandcovariance,whenusedalone(i.e.,
withoutthetermfunctionresultintoasinglenumber
CovarianceandCorrelationMatrices
r.R
.B
.G
ho
ng
ad
e
Canbeappliedtomultivariatedatawheremultipleresponses,
orobservations,areobtainedfromasingleprocess
Thecovarianceandcorrelationmatricesassumethatthe
multivariatedataarearrangedinamatrixwherethecolumns
aredifferentvariablesandtherowsaredifferentobservations
ofthosevariables
Insignalprocessing,therowsarethewaveformtimesamples,
andthecolumnsarethedifferentsignalchannelsor
observationsofthesignal
Thecovariancematrixgivesthevarianceofthecolumnsofthe
datamatrixinthediagonalswhilethecovariancebetween
columnsisgivenbytheoffdiagonals
ad
e
Correlationmatrixisrelatedtothecovariancematrixbythe
equation:
r.R
.B
.G
ho
ng
Thecorrelationmatrixisasetofcorrelationcoefficients
betweenwaveformobservationsorchannelsandhasasimilar
positionalrelationshipasinthecovariancematrix
Sincethediagonalsinthecorrelationmatrixgivethe
correlationofagivenvariableorwaveformwithitself,theywill
allequal1(rxx(0)=1),andtheoffdiagonalswillvarybetween
1
SAMPLINGTHEORYANDFINITEDATA
CONSIDERATIONS
ad
e
Toconvertananalogwaveformintoadigitizedversion:
ho
ng
samplingthewaveformatdiscretepointsintime
ifthewaveformislongerthanthecomputermemory,isolatinga
segmentoftheanalogwaveformfortheconversion(windowing)
r.R
.B
.G
TheShannonSamplingTheoremstatesthatanysinusoidal
waveformcanbeuniquelyreconstructedprovideditis
sampledatleasttwiceinoneperiod
Thesamplingfrequency,fs,mustbe2fsinusoid
ShannonsSamplingTheoremstatesthatacontinuous
waveformcanbereconstructedwithoutlossofinformation
providedthesamplingfrequencyisgreaterthantwicethe
highestfrequencyintheanalogwaveform:
Samplingfunction
r.R
.B
.G
ho
ng
ad
e
Thesamplingprocessisequivalenttomultiplyingtheanalog
waveformbyarepeatingseriesofshortpulses
Thisrepeatingseriesofshortpulsesissometimesreferredto
asthesamplingfunction
Thesamplingfunctioncanbestatedmathematicallyusingthe
impulseresponse
whereTs isthesampleintervalandequals1/fs
Forananalogwaveform,x(t),thesampledversion,x(n),is
givenbymultiplyingx(t)bythesamplingfunction:
Effectsofsampling
Multiplicationinthetime
domainisequivalentto
convolutioninfrequency
domain(andviceversa)
Hence,thefrequency
characteristicofasampled
waveformisjustthe
convolutionoftheanalog
waveformspectrumwith
thesamplingfunction
spectrum
Itwouldbepossibleto
recovertheoriginal
spectrumsimplybyfiltering
thesampleddatabyan
ideallowpassfilterwitha
bandwidth>fmax
r.R
.B
.G
ho
ng
ad
e
CONV
Aliasing
Spectrumthatresultsifthedigitizeddataweresampledatfs <2fmax,in
thiscasefs =1.5fmax
r.R
.B
.G
ho
ng
ad
e
Thereflectedportionofthespectrumhas
becomeintermixedwiththeoriginal
spectrum,andnofiltercanunmix them
Whenfs <2fmax,thesampleddatasuffers
fromspectraloverlap,better knownas
aliasing
Thesampleddatanolongerprovidesa
uniquerepresentationoftheanalog
waveform,andrecoveryisnotpossible
Aliasingmustbeavoidedeitherby:
useofveryhighsamplingratesrates
thatarewellabovethebandwidthofthe
analogsystem
orbyfilteringtheanalogsignalbefore
analogtodigitalconversion
Antialiasingfilters
Example
ng
ad
e
AnECGsignalof1voltpeaktopeakhasabandwidthof0.01to100Hz.
Assumethatbroadbandnoisemaybepresentinthesignalatabout0.1volts
(i.e.,20db belowthenominalsignallevel).Thissignalisfilteredusingafour
polelowpassfilter.Whatsamplingfrequencyisrequiredtoensurethatthe
errorduetoaliasingislessthan60db (0.001volts)?
r.R
.B
.G
ho
Thenoiseatthesamplingfrequencymustbereducedanother40db (20*log
(0.1/0.001))bythefourpolefilter.Afourpolefilterwitha
cutoffof100Hz(requiredtomeetthefidelityrequirementsoftheECGsignal)
wouldattenuatethewaveformatarateof80db perdecade.Forafourpole
filtertheasymptoticattenuationisgivenas:
Attenuation=80log(f2/fc)db
Toachievetherequiredadditional40db ofattenuationrequiredbythe
problemfromafourpolefilter:
80log(f2/fc)=40log(f2/fc)=40/80=0.5
f2/fc=10.5=;f2=3.16 100=316Hz
Thustomeetthesamplingcriterion,thesamplingfrequencymustbeat
least632Hz,twicethefrequencyatwhichthenoiseisadequatelyattenuated
Unfortunately,inorderforthisimpulsefunctiontoproduceanidealfilter,
itmustbeinfinitelylong
Howeveriffs>>fmax,asisoftenthecase,thenanyreasonablelowpass
filterwouldsufficetorecovertheoriginalwaveform
r.R
.B
.G
ho
ng
ad
e
Recoveryofawaveformwhenthesamplingfrequencyismuchmuch
greaterthattwicethehighestfrequencyinthesampledwaveform(fs=
10fmax)iseasierwithpracticalLPF
Inthiscase,thelowpassfilter(dottedline)neednothaveassharpa
cutoff.
EdgeEffects
r.R
.B
.G
ho
ng
ad
e
Advantageofdealingwithinfinitedataisthatone
neednotbeconcernedwiththeendpoints
Finitedataconsistofnumericalsequenceshaving
afixedlengthwithfixedendpointsatthe
beginningandendofthesequence
Someoperations,suchasconvolution,may
produceadditionaldatapointswhilesome
operationswillrequireadditionaldatapointsto
completetheiroperationonthedataset
Howtoaddoreliminatedatapoints?
Extendingdatalength
ad
e
Threecommonstrategiesforextendingadatasetwhen
additionalpointsareneeded:
.G
ho
ng
extendingwithzeros(oraconstant),termedzeropadding;
extendingusingperiodicityorwraparound;
extendingbyreflection,alsoknownassymmetricextension
r.R
.B
Inthezeropaddingapproach,zerosareaddedtotheendor
beginningofthedatasequence
Thisapproachisfrequentlyusedinspectralanalysisandis
justifiedbytheimplicitassumptionthatthewaveformiszero
outsideofthesampleperiodanyway
Avariantofzeropaddingisconstantpadding,wherethedata
sequenceisextendedusingaconstantvalue,oftenthelast(or
first)valueinthesequence
r.R
.B
.G
ho
ng
ad
e
Ifthewaveformcanbereasonablythoughtofasone
cycleofaperiodicfunction,thenthewraparound
approachisclearlyjustifieddataareextendedby
tackingontheinitialdatasequencetotheendofthe
datasetandvisaversa
Thisisquiteeasytoimplementnumerically:simply
makealloperationsinvolvingthedatasequenceindex
moduloN,whereNistheinitiallengthofthedataset
Thesetwoapproacheswill,ingeneral,producea
discontinuityatthebeginningorendofthedataset,
whichcanleadtoartifactincertainsituations
Thesymmetricreflectionapproacheliminatesthis
discontinuitybytakingontheendpointsinreverse
order(orbeginningpointsifextendingthebeginningof
thedatasequence)
ad
e
ng
ho
.G
r.R
.B
SpectralAnalysis:ClassicalMethods
.G
r.R
.B
ho
ng
Manybiologicalsignalsdemonstrateinterestingordiagnosticallyusefulproperties
whenviewedinthesocalledfrequencydomain,E.g.heartrate,EMG,EEG,ECG,
eyemovementsandothermotorresponses,acousticheartsounds,andstomach
andintestinalsounds
Determiningthefrequencycontentofawaveformistermedspectralanalysis
Methodscanbedividedintotwobroadcategories
classicalmethodsbasedontheFouriertransform
modernmethodssuchasthosebasedontheestimationofmodelparameters
Theaccuratedeterminationofthewaveformsspectrumrequiresthatthesignal
beperiodic,oroffinitelength,andnoisefree
Butmanybiologicalsignalsare
eitherinfiniteorofsufficientlengththatonlyaportionofitisavailablefor
analysis
oftencorruptedbysubstantialamountsofnoiseorartifact
Allspectralanalysistechniquesmustnecessarilybeapproximate;theyare
estimatesofthetruespectrum
Thevariousspectralanalysisapproachesattempttoimprovetheestimation
accuracyofspecificspectralfeatures
ad
e
Twospectralfeaturesofpotentialinterestare:
ad
e
theoverallshapeofthespectrum,termedthespectralestimate,
and/or
localfeaturesofthespectrumsometimesreferredtoasparametric
estimates
r.R
.B
.G
ho
ng
Techniquesthatprovidegoodspectralestimationarepoor
localestimatorsandviceversa
THEFOURIERTRANSFORM:FOURIERSERIESANALYSIS
ad
e
ng
ho
.G
r.R
.B
ClassicalFouriertransform(FT)methodisthemoststraightforwardforspectral
estimate
Anyperiodic waveformcanberepresentedbyaseriesofsinusoidsthatareatthe
samefrequencyas,ormultiplesof,thewaveformfrequency
Ifawaveformcanbebrokendownintoaseriesofsines orcosinesofdifferent
frequencies,theamplitudeofthesesinusoidsmustbeproportionaltothe
frequencycomponentcontainedinthewaveformatthosefrequencies
Considerthecasewheresines andcosinesareusedtorepresentthefrequency
components:tofindtheappropriateamplitudeofthesecomponentsitisonly
necessarytocorrelate(i.e.,multiply)thewaveformwiththesineandcosinefamily,
andaverage(i.e.,integrate)overthecompletewaveform(oroneperiodifthe
waveformisperiodic)
whereT istheperiodortimelengthofthe
waveform,fT =1/T,andm issetof
integers,possiblyinfinite:m=1,2,3,...,
definingthefamilymember.
Thisgivesrisetoafamilyofsines andcosines
havingharmonicallyrelatedfrequencies,
mfT
r.R
.B
.G
ho
ng
ad
e
Fourierseriesanalysisusesaprobingfunctioninwhichthefamily
consistsofharmonicallyrelatedsinusoids
Thesines andcosinesinthisfamilyhavevalidfrequenciesonlyat
valuesofm/T,whichiseitherthesamefrequencyasthewaveform
(whenm=1)orhighermultiples(whenm>1)thataretermed
harmonics
Sincethisapproachrepresentswaveformsbyharmonicallyrelated
sinusoids,theapproachissometimesreferredtoasharmonic
decomposition
Forperiodicfunctions,theFouriertransformandFourierseries
constituteabilateraltransform:theFouriertransformcanbe
appliedtoawaveformtogetthesinusoidalcomponentsandthe
Fourierseriessineandcosinecomponentscanbesummedto
reconstructtheoriginalwaveform:
.(1)
ad
e
ng
ho
.G
r.R
.B
D
Twoperiodicfunctionsandtheirapproximationsconstructedfroma
limitedseriesofsinusoids.
Uppergraphs:Asquarewaveisapproximatedbyaseriesof3and6sinewaves.
Lowergraphs:Atrianglewaveisapproximatedbyaseriesof3and6cosine
waves
r.R
.B
.G
ho
ng
ad
e
Spectralinformationisusuallypresentedasafrequencyplot,
aplotofsineandcosineamplitudevs.componentnumber,or
theequivalentfrequency
Toconvertfromcomponentnumber,m,tofrequency,f,note
thatf=m/T,whereTistheperiodofthefundamental.
Indigitizedsignals,thesamplingfrequencycanalsobeused
todeterminethespectralfrequency
Ratherthanplotsineandcosineamplitudes,itismore
intuitivetoplottheamplitudeandphaseangleofasinusoidal
waveusingtherectangulartopolartransformation
ad
e
ng
ho
.G
D
r.R
.B
Atriangleorsawtooth wave(left)andthefirst10termsofitsFourier
series(right)
Notethatthetermsbecomequitesmallafterthesecondterm
Symmetry
r.R
.B
.G
ho
ng
ad
e
Somewaveformsaresymmetricalorantisymmetricalaboutt
=0,sothatoneortheotherofthecomponents,a(k)orb(k)in
Eq.(1),willbezero
Ifthewaveformhasmirrorsymmetryaboutt=0,thatis,x(t)
=x(t),thenmultiplicationsbyasinefunctionswillbezero
irrespectiveofthefrequency,andthiswillcauseallb(k)terms
tobezeros
Suchmirrorsymmetryfunctionsaretermedeven functions
Ifthefunctionhasantisymmetry,x(t)=x(t),asocalledodd
function,thenallmultiplicationswithcosinesofany
frequencywillbezero,causingalla(k)coefficientstobezero
Functionsthathavehalfwavesymmetrywillhavenoeven
coefficients,andbotha(k)andb(k)willbezeroforevenm
Thesesymmetriesareusefulforreducingthecomplexityof
solvingforthecoefficientswhensuchcomputationsaredone
manually
r.R
.B
.G
ho
ng
ad
e
FunctionSymmetries
RectangularWindow
r.R
.B
.G
ho
ng
ad
e
Thedigitizedwaveformmustnecessarilybetruncatedatleast
tothelengthofthememorystoragearray
Thisprocessiscalledwindowing
Thewindowingprocesscanbethoughtofasmultiplyingthe
databysomewindowshape
Ifthewaveformissimplytruncatedandnofurthershapingis
performedontheresultantdigitizedwaveform(asisoftenthe
case),thenthewindowshapeisrectangular bydefault
Othershapescanbeimposedonthedatabymultiplyingthe
digitizedwaveformbythedesiredshape
Windowingcreatessomeeffects(tobediscussedlater!)
AnotherrepresentationofFourierseries
TheequationsforcomputingFourierseriesanalysisofdigitizeddataare
thesameasforcontinuousdataexcepttheintegrationisreplacedby
summation
Equationsarepresentedusingcomplexvariablesnotationsothatboththe
sineandcosinetermscanberepresentedbyasingleexponentialterm
usingEulersidentity
ho
DiscreteFouriertransformbecomes:
r.R
.B
.G
ng
ad
e
TheinverseFouriertransformcanbecalculatedas:
givesthemagnitudeforthesinusoidalrepresentation
oftheFourierserieswhiletheangleofX(m)givesthephase
angleforthisrepresentation,sinceX(m)canalsobewrittenas
ng
ad
e
r.R
.B
.G
ho
ThediscreteFouriertransformproducesafunctionofm
Toconvertthistofrequencynotethat:
TheequationforthediscreteFouriertransformcanalsobe
writtenas:
FourierTransformForAperiodicFunctions
r.R
.B
.G
ho
ng
ad
e
Ifthefunctionisnotperiodic,itcanstillbe
accuratelydecomposedintosinusoidsifitis
aperiodic;thatis,itexistsonly forawelldefined
periodoftime,andthattimeperiodisfully
representedbythedigitizedwaveform
Thesinusoidalcomponentscanexistatall
frequencies,notjustmultiplefrequenciesor
harmonics
Theanalysisprocedureisthesameasforaperiodic
function,exceptthatthefrequenciesobtainedare
reallyonlysamplesalongacontinuousfrequency
spectrum
ad
e
ng
ho
.G
r.R
.B
Thefrequency
spectrumofa
periodictriangle
waveforthree
differentperiods
Astheperiod
getslonger,
approachingan
aperiodic
function,the
spectralshape
doesnotchange,
butthepointsget
closertogether
FrequencyResolution
r.R
.B
.G
ho
ng
ad
e
FromthediscreteFourierseriesequation,thenumberofpoints
producedbytheoperationisN,thenumberofpointsinthedata
set
Sincethespectrumproducedissymmetricalaboutthemidpoint,
N/2(orfs/2infrequency),onlyhalfthepointscontainunique
information
IfthesamplingtimeisTs,theneachpointinthespectrarepresents
afrequencyincrementof1/(NTs)
Asaroughapproximation,thefrequencyresolutionofthespectra
willbethesameasthefrequencyspacing,1/(NTs)
FrequencyspacingofthespectrumproducedbytheFourier
transformcanbedecreasedbyincreasingthelengthofthedata,N
Increasingthesampleinterval,Ts,shouldalsoimprovethe
frequencyresolution,butsincethatmeansadecreaseinfs,the
maximumfrequencyinthespectra,fs /2isreducedlimitingthe
spectralrange
OnesimplewayofincreasingNevenafterthewaveformhasbeen
sampledistousezeropadding
TruncatedFourierAnalysis:DataWindowing
r.R
.B
Rectangular
Barlett (Triangular)
Hamming
Hanning
TruncatedGaussian
.G
ho
ng
ad
e
Often,awaveformisneitherperiodicoraperiodic,buta
segmentofamuchlongerpossiblyinfinitetimeseries(E.g.
ECG)
Onlyaportionofsuchwaveformscanberepresentedinthe
finitememoryofthecomputer,andsomeattentionmustbe
paidtohowthewaveformistruncated
Typesofwindowingused:
r.R
.B
.G
ho
ng
ad
e
WindowFunctions(a.k.a.TaperingFunctions)
ad
e
Rectangular
.G
ho
ng
Barlett (Triangular)
Hanning
r.R
.B
Hamming
Gaussian
Effectsofwindowing
r.R
.B
.G
ho
ng
ad
e
Whenadatasetiswindowed,whichisessentialifthedatasetis
largerthanthememorystorage,thenthefrequency characteristics
ofthewindowbecomepartofthespectralresult
Thus allwindowsproducetwotypesofartifact
Theactualspectrumiswidenedbyanartifacttermedthemainlobe,
andadditionalpeaksaregeneratedtermedthesidelobes
ad
e
ng
ho
.G
r.R
.B
SelectingtheWindowFunction
r.R
.B
.G
ho
ng
ad
e
Selectingtheappropriatewindow,dependsonwhatspectral
featuresareofinterest
Ifthetaskistoresolvetwonarrowbandsignalscloselyspacedin
frequency,thenawindowwiththenarrowestmainlobe (the
rectangularwindow)ispreferred
Ifthereisastrongandaweaksignalspacedamoderatedistance
apart,thenawindowwithrapidlydecayingsidelobes ispreferredto
preventthesidelobes ofthestrongsignalfromoverpoweringthe
weaksignal
Iftherearetwomoderatestrengthsignals,onecloseandtheother
moredistantfromaweaksignal,thenacompromisewindowwith
amoderatelynarrowmainlobe andamoderatedecayinsidelobes
couldbethebestchoice
Oftenthemostappropriatewindowisselectedbytrialanderror
PowerSpectrum
r.R
.B
.G
ho
ng
ad
e
ThepowerspectrumiscommonlydefinedastheFourier
transformoftheautocorrelationfunction
Incontinuousanddiscretenotation,thepowerspectrum
equationbecomes:
Sincetheautocorrelationfunctionhasoddsymmetry,thesine
terms,b(k)willallbezero
PowerSpectrum(DirectApproach)
Thedirectapproachismotivatedbythefactthattheenergycontainedin
ananalogsignal,x(t),isrelatedtothemagnitudeofthesignalsquared,
integratedovertime
ByanextensionofParsevals theoremitiseasytoshowthat
.G
ho
ng
ad
e
r.R
.B
Hence equalstheenergydensityfunctionoverfrequency,also
referredtoastheenergyspectraldensity,thepowerspectraldensity,or
simplythepowerspectrum
Inthedirectapproach,thepowerspectrumiscalculatedasthe
magnitudesquaredoftheFouriertransformofthewaveformofinterest:
Parsevals Theorem:Thesum(orintegral)ofthesquareofafunctionis
equaltothesum(orintegral)ofthesquareofitstransform
Periodogram
r.R
.B
.G
ho
ng
ad
e
Whilethepowerspectrumcanbeevaluatedbyapplyingthe
FFTtotheentirewaveform,averagingisoftenused,
particularlywhentheavailablewaveformisonlyasampleofa
longersignal
Insuchverycommonsituations,powerspectrumevaluation
isnecessarilyanestimationprocess,andaveragingimproves
thestatisticalpropertiesoftheresult
Whenthepowerspectrumisbasedonadirectapplicationof
theFouriertransformfollowed byaveraging,itiscommonly
referredtoasanaverageperiodogram
Selectionofdatawindowandaveragingstrategyisusually
basedonexperimentationwiththeactualdata
Averagingisusuallyachievedbydividingthewaveformintoa
numberofsegments,possiblyoverlapping,andevaluatingthe
Fouriertransformoneachofthesesegments
Welchmethodofspectralanalysis
r.R
.B
.G
ho
ng
ad
e
Oneofthemostpopularprocedurestoevaluatetheaverage
periodogram isattributedtoWelchandisamodificationof
thesegmentationschemeoriginallydevelopedbyBartlett
Inthisapproach,overlappingsegmentsareused,anda
windowisappliedtoeachsegment
Byoverlappingsegments,moresegmentscanbeaveragedfor
agivensegmentanddatalength
Averagedperiodograms obtainedfromnoisydatatraditionally
averagespectrafromhalfoverlappingsegments;thatis,
segmentsthatoverlapby50%.
ad
e
ng
ho
.G
r.R
.B
D
Awaveformisdividedintothreesegmentswitha50%overlapbetweeneach
segment
IntheWelchmethodofspectralanalysis,theFouriertransformofeach
segmentwouldbecomputedseparately,andanaverageofthethree
transformswouldprovidetheoutput
DigitalFilters
r.R
.B
.G
ho
ng
ad
e
Filtersarecloselyrelatedtospectralanalysissincethegoalof
filteringistoreshapethespectrumtoonesadvantage
Mostnoiseisbroadband(thebroadestbandnoisebeing
whitenoisewithaflatspectrum)andmostsignalsare
narrowband;hence,filtersthatappropriatelyreshapea
waveformsspectrumwillalmostalwaysprovidesome
improvement inSNR
Abasicfiltercanbeviewedasalinearprocessinwhichthe
inputsignalsspectrumisreshapedinsomewelldefined
manner
Filtersdifferinthewaytheyachievethisspectralreshaping,
andcanbeclassifiedintotwogroupsbasedontheir
approach:
finiteimpulseresponse(FIR)filters
infiniteimpulseresponse(IIR)filters
THEZTRANSFORM
r.R
.B
.G
ho
ng
ad
e
Frequencybasedanalysisintroducedinthelastchapterisamost
usefultoolforanalyzingsystemscannotbeappliedtotransient
responsesofinfinitelength,suchasstepfunctions,orsystemswith
nonzeroinitialconditions
MotivatedthedevelopmentoftheLaplacetransformintheanalog
domain
Laplaceanalysisusesthecomplexvariables (s=+j)asa
representationofcomplexfrequencyinplaceofj intheFourier
transform
TheZtransformisadigitaloperationanalogoustotheLaplace
transformintheanalogdomain,anditisusedinasimilarmanner
TheZtransformisbasedaroundthecomplexvariable,z,wherez is
anarbitrarycomplexnumber, ej
Thisvariableisalsotermedthecomplexfrequency,andaswithits
timedomaincounterpart,theLaplacevariables,itispossibleto
substituteej forz toperformastrictlysinusoidalanalysis
If
issetto1,thenz=ej.Thisiscalledevaluatingz ontheunitcircle
TheZtransform(similartotheFouriertransformequation)is:
ad
e
wherez =anarbitrarycomplexvariable
r.R
.B
.G
ho
ng
Probingfunctionforthistransformissimplyzn
Inanyrealapplication,thelimitofthesummationwillbe
finite,usuallythelengthofx(n)
Whenidentifiedwithadatasequence,suchasx(n) above,zn
representsanintervalshiftofn samples,oranassociated
timeshiftofnTs seconds
Thistimeshiftingpropertyofzn canbeformallystatedas:
ThetimeshiftingcharacteristicoftheZ
transformcanbeusedtodefineaunit
delayprocess,z1
DigitalTransferFunction
r.R
.B
.G
ho
ng
ad
e
MostusefulapplicationsoftheZtransformliesinitsabilityto
definethedigitalequivalentofatransferfunction
Byanalogytolinearsystemanalysis,thedigitaltransfer
functionisdefinedas:
Unlikeanalogsystems,theorderofthenumerator,N,need
notbelessthan,orequalto,theorderofthedenominator,D,
forstability
Infact,systemsthathaveadenominatororderof1aremore
stablethatthosehavinghigherorderdenominators
ng
ad
e
Fromthedigitaltransferfunction,H(z),itispossibleto
determinetheoutputgivenanyinput
r.R
.B
.G
ho
Theinputoutputordifferenceequationanalogoustothe
timedomainequationandcanbeobtainedbyapplyingthe
timeshiftinterpretationtothetermzn
equationassumesthata(0)=1
Filterdesign,then,issimplythedeterminationofthe
appropriatefiltercoefficients,a(n)andb(n),thatprovidethe
desiredspectralshaping
ho
ng
ad
e
IfthefrequencyspectrumofH(z) isdesired,itcanbe
obtainedfromamodificationsubstitutingz=ej as:
r.R
.B
.G
Frequencycanbeobtainedfromthevariablem bymultiplying
byfs/N or1/(NTs)
FINITEIMPULSERESPONSE(FIR)FILTERS
r.R
.B
.G
ho
ng
ad
e
FIRfiltershavetransferfunctionsthathaveonlynumerator
coefficients, i.e.,H(z)=B(z)
Thisleadstoanimpulseresponsethatisfinite
Merits:
stable
linearphaseshifts
haveinitialtransientsthatareoffinitedurations
theirextensionto2dimensionalapplicationsisstraightforward
Demerits:
lessefficientintermsofcomputertimeandmemory
FIRfiltersarealsoreferredtoasnonrecursive becauseonly
theinput(nottheoutput)isusedinthefilteralgorithm
FIRfilteringhasalsobeenreferredtoasamoving
average process
ThegeneralequationforanFIRfilteris:
Similartoconvolution
ad
e
whereb(n) isthecoefficientfunction(alsoreferredtoastheweighting
function)oflengthL,x(n) istheinput,andy(n) istheoutput
r.R
.B
.G
ho
ng
Filtercoefficients(orweights)ofanFIRfilterarethesameasthe
impulseresponseofthefilter
Sincethefrequencyresponseofaprocesshavinganimpulse
responseh(n) issimplytheFouriertransformofh(n),thefrequency
responseofanFIRfilterhavingcoefficientsb(n) isjusttheFourier
transformofb(n):
Theinverseoperation,goingfromadesiredfrequencyresponseto
thecoefficientfunction,b(n),isknownasfilterdesign
SincethefrequencyresponseistheFouriertransformofthefilter
coefficients,thecoefficientscanbefoundfromtheinverseFourier
transformofthedesiredfrequencyresponse
FIRFilterDesign
r.R
.B
.G
ho
ng
ad
e
r.R
.B
.G
ho
ng
FFTofthisfunctionissameas
animpulseresponse
Thiscoefficientfunctionmust
beinfinitelylongtoproduce
thefiltercharacteristicsofan
idealfilter
Truncatingitwillresultina
lowpass filterthatislessthan
ideal
ad
e
Filterfunction
r.R
.B
.G
ho
ng
ad
e
Effectsofb(n) truncationwithrectangular
window
Theweightingfunctionswereabruptlytruncatedat17and65
coefficients(rectangularwindow)
Theartifactsassociatedwiththistruncationareclearlyseen
Thelowpass cutofffrequencyis100Hz
r.R
.B
.G
ho
ng
ad
e
Effectsofb(n) truncationwith
Hammingwindow
Theovershootinthepassband hasdisappearedandtheoscillations
arebarelyvisibleintheplot
Highpass,Bandpass,andBandstop
filters
r.R
.B
.G
ho
ng
ad
e
Arederivedinthesamemannerfromequationsgeneratedby
applyinganinverseFTtorectangularstructureshavingthe
appropriateassociatedshape
ad
e
ng
r.R
.B
.G
ho
MATLABDemoLPFFIRDesign
StepsforDesigningFIRFilters
r.R
.B
.G
ho
ng
ad
e
DerivativeOperation
r.R
.B
.G
ho
ng
ad
e
Thederivativeisacommonoperationinsignalprocessingandis
particularlyusefulinanalyzingcertainphysiologicalsignals
Digitaldifferentiationisdefinedasx/t andcanbeimplemented
bytakingthedifferencebetweentwoadjacentpoints,scalingby
1/Ts,andrepeatingthisoperationalongtheentirewaveform
AsFIRfilterthisisequivalenttoatwocoefficientfilter,[1,
+1]/Ts,
Thefrequencycharacteristicofthederivativeoperationisalinear
increasewithfrequencysothereis
Considerablegainatthehigherfrequencies
Sincethehigherfrequenciesfrequentlycontainagreater
percentageofnoise,thisoperationtendstoproduceanoisy
derivativecurvehenceweusetwopointcentraldifference
algorithm
TheTwoPointCentralDifferenceAlgorithm
ho
ng
ad
e
Thetwopointcentraldifferencealgorithmusestwo
coefficientsofequalbutoppositevaluespacedL pointsapart,
asdefinedbytheinputoutputequation:
r.R
.B
.G
whereL istheskipfactorthatinfluencestheeffectivebandwidth,and
Ts isthesampleinterval
Thefiltercoefficientsforthetwopointcentraldifference
algorithmwouldbe:
Frequencycharacteristicofthederivative
operation
ad
e
Ideal
r.R
.B
.G
ho
ng
FIRimplementation
ad
e
ng
ho
.G
r.R
.B
D
(A)Thederivativewascalculatedbytakingthedifferencein
adjacentpointsandscalingbythesamplefrequency.
(B)Thederivativewascomputedusingthetwopointcentral
differencealgorithmwithaskipfactorof4
TimeFrequencyAnalysis
ad
e
ng
ho
.G
Spectralanalysistechniquesdevelopedthusfarrepresentpowerfulsignal
processingtoolsifoneisnotespeciallyconcernedwithsignaltiming
Classicalormodernspectralmethodsprovideacompleteandappropriate
solutionforwaveformsthatarestationary;thatis,waveformsthatdonot
changeintheirbasicpropertiesoverthelengthoftheanalysis
Manywaveformsparticularlythoseofbiologicaloriginarenot
stationary,andchangesubstantiallyintheirpropertiesovertime
Fourieranalysisprovidesagooddescriptionofthefrequenciesina
waveform,butnottheirtiming
Timingisencodedinthephaseportionofthetransform,andthis
encodingisdifficulttointerpretandrecover
IntheFouriertransform,specificeventsintimearedistributedacrossall
ofthephasecomponents
Alocalfeatureintimehasbeentransformedintoaglobalfeaturein
phase
Timinginformationisoftenofprimaryinterestinmanybiomedicalsignals,
andthisisalsotrueformedicalimageswheretheanalogousinformation
islocalizedinspace
r.R
.B
Methods
r.R
.B
.G
ho
ng
ad
e
Awiderangeofapproacheshavebeen
developedtotrytoextractbothtimeand
frequencyinformationfromawaveform
Basicallytheycanbedividedintotwogroups:
timefrequencymethods
timescalemethods(waveletanalysis)
ShortTermFourierTransform:The
Spectrogram
r.R
.B
.G
ho
ng
ad
e
Thefirsttimefrequencymethodswerebasedonthe
straightforwardapproachofslicingthewaveformof
interestintoanumberofshortsegmentsandperforming
theanalysisoneachofthesesegments,usuallyusingthe
standardFouriertransform
Awindowfunctionisappliedtoasegmentofdata,
effectivelyisolatingthatsegmentfromtheoverall
waveform,andtheFouriertransformisappliedtothat
segment
Thisistermedthespectrogram orshorttermFourier
transform(STFT)sincetheFourierTransformisappliedto
asegmentofdatathatisshorter,oftenmuchshorter,than
theoverallwaveform
Selectingthemostappropriatewindowlengthcanbe
critical
ad
e
Thebasicequationforthespectrograminthecontinuous
domainis:
.G
r.R
.B
Thediscreteversion
ho
ng
Therearetwomainproblemswiththespectrogram:
(1)selectinganoptimalwindowlengthfordatasegments
thatcontainseveraldifferentfeaturesmaynotbepossible,
(2)thetimefrequencytradeoff:shorteningthedatalength,
N,toimprovetimeresolutionwillreducefrequency
resolutionwhichisapproximately1/(NTs)
ho
ng
ad
e
Ifwindowismadesmallertoimprovethetimeresolution,
thenthefrequencyresolutionisdegradedandvisaversa
Thistimefrequencytradeoffhasbeenequatedtoan
uncertaintyprinciplewheretheproductoffrequency
resolution(expressedasbandwidth,B)andtime,T,mustbe
greaterthansomeminimum
r.R
.B
.G
STFThasbeenusedsuccessfullyinawidevarietyofproblems,
particularlythosewhereonlyhighfrequencycomponentsare
ofinterestandfrequencyresolutionisnotcritical
Theareaofspeechprocessinghasbenefittedconsiderably
fromtheapplicationoftheSTFT
Whereappropriate,theSTFTisasimplesolutionthatrestson
awellunderstoodclassicaltheory(i.e.,theFouriertransform)
andiseasytointerpret
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
ad
e
ng
ho
.G
r.R
.B
WaveletAnalysis
r.R
.B
.G
ho
ng
ad
e
InabilityoftheFouriertransformtodescribe
bothtimeandfrequencycharacteristicsofthe
waveformledtoanumberofdifferent
approaches
Thewavelettransformcanbeusedasyet
anotherwaytodescribethepropertiesofa
waveformthatchangesovertime,butinthis
casethewaveformisdividednotintosections
oftime,butsegmentsofscale
THECONTINUOUSWAVELETTRANSFORM
ho
ng
Avarietyofdifferentprobingfunctionsmaybeused,butthefamilyalways
consistsofenlargedorcompressedversionsofthebasicfunction,aswell
astranslations
Continuouswavelettransform(CWT)equationisdefinedas:
ad
e
r.R
.B
.G
r.R
.B
.G
ho
ng
ad
e
ThewaveletshownisthepopularMorlet wavelet,namedaftera
pioneerofwaveletanalysis,andisdefinedbytheequation
r.R
.B
.G
ho
ng
ad
e
Waveletcoefficients,W(a,b),describethecorrelation
betweenthewaveformandthewaveletatvarious
translationsandscales:thesimilaritybetweenthewaveform
andthewaveletatagivencombinationofscaleandposition,
a,b
Coefficientsprovidetheamplitudesofaseriesofwavelets,
overarangeofscalesandtranslations,thatwouldneedtobe
addedtogethertoreconstructtheoriginalsignal
Waveletanalysiscanbethoughtofasasearchoverthe
waveformofinterestforactivitythatmostclearly
approximatestheshapeofthewavelet
Thissearchiscarriedoutoverarangeofwaveletsizes:the
timespanofthewaveletvariesalthoughitsshaperemains
thesame
Waveletcoefficientsrespondtochangesinthewaveform,
morestronglytochangesonthesamescaleasthewavelet,
andmoststrongly,tochangesthatresemblethewavelet
r.R
.B
.G
ho
ng
ad
e
Ifthewaveletfunction,(t),isappropriatelychosen,thenitis
possibletoreconstructtheoriginalwaveformfromthe
waveletcoefficientsjustasintheFouriertransform
AsCWTdecomposesthewaveformintocoefficientsoftwo
variables,a andb,adoublesummation(orintegration)is
requiredtorecovertheoriginalsignalfromthecoefficients
WaveletTimeFrequencyCharacteristics
ng
Waveletsprovideacompromiseinthebattlebetweentimeandfrequency
localization:theyarewelllocalizedinbothtimeandfrequency,butnot
preciselylocalizedineither
Measureofthetimerangeofaspecificwavelet,t,canbespecifiedby
thesquarerootofthesecondmomentofagivenwaveletaboutitstime
center(i.e.,itsfirstmoment)
ad
e
r.R
.B
.G
ho
wheret0 isthecentertime,orfirstmomentof
thewavelet
Inmathematics,amomentis,looselyspeaking,aquantitativemeasureoftheshapeofa
setofpoints.The"secondmoment",forexample,iswidelyusedandmeasuresthe
"width"(inaparticularsense)ofasetofpointsinonedimensionorinhigherdimensions
measurestheshapeofacloudofpointsasitcouldbefitbyanellipsoid.Othermoments
describeotheraspectsofadistributionsuchashowthedistributionisskewedfromits
mean,orpeaked.Anydistributioncanbecharacterizedbyanumberoffeatures(suchas
themean,thevariance,theskewness,etc.),andthemomentsofafunctiondescribethe
natureofitsdistribution
ng
ad
e
Similarlythefrequencyrange,,isgivenby:
r.R
.B
.G
ho
where() isthefrequencydomainrepresentation(i.e.,Fourier
transform)of(t/a),and0 isthecenterfrequencyof()
Thetimeandfrequencyrangesofagivenfamilycanbe
obtainedfromthemotherwavelet
r.R
.B
.G
ho
ng
ad
e
ForMexicanhatwavelet
ho
ng
ad
e
Thefrequencyrange,orbandwidth,wouldbetherangeofthemother
Waveletdividedbya:
(a)=/
Ifwemultiplythefrequencyrangebythetimerange,theascanceland
weareleftwithaconstantthatistheproductoftheconstants
Productoftherangesisinvarianttodilationandthattherangesareinversely
related;increasingthefrequencyrange,(a),decreasesthetimerange,
t(a)
TheserangescorrelatetothetimeandfrequencyresolutionoftheCWT
Decreasingthewavelettimerange(bydecreasinga)providesamore
accurateassessmentoftimecharacteristics(i.e.,theabilitytoseparateout
closeeventsintime)attheexpenseoffrequencyresolution,andviceversa
CWTwillprovidebetterfrequencyresolutionwhena islargeandthelength
ofthewavelet(anditseffectivetimewindow)islong
Conversely,whena issmall,thewaveletisshortandthetimeresolutionis
maximum,butthewaveletonlyrespondstohighfrequencycomponents
r.R
.B
.G
ng
ad
e
x(t)
r.R
.B
.G
ho
CWT
Mother
wavelet
ad
e
ng
ho
r.R
.B
.G
CWTrepresentationsallow
detectingthefiducial points
forECG
THEDISCRETEWAVELETTRANSFORM
r.R
.B
.G
ho
ng
ad
e
TheCWThasoneseriousproblem:itishighlyredundant.
TheCWTprovidesanoversamplingoftheoriginalwaveform:many
morecoefficientsaregeneratedthanareactuallyneededto
uniquelyspecifythesignal
Willbecostlyiftheapplicationcallsforrecoveryoftheoriginal
signal
Forrecovery,allofthecoefficientswillberequiredandthe
computationaleffortcouldbeexcessive
Inapplicationsthatrequirebilateraltransformations,wewould
preferatransformthatproducestheminimumnumberof
coefficientsrequiredtorecoveraccuratelytheoriginalsignal
Thediscretewavelettransform(DWT)achievesthisbyrestricting
thevariationintranslationandscale,usuallytopowersof2
Whenthescaleischangedinpowersof2,thediscretewavelet
transformissometimestermedthedyadicwavelettransform
TheDWTisoftenintroducedintermsofitsrecoverytransform
ad
e
.G
ho
ng
r.R
.B
Newconceptisintroducedtermedthescalingfunction,afunctionthat
facilitatescomputationoftheDWT
ToimplementtheDWTefficiently,thefinestresolutioniscomputedfirst
Thecomputationthenproceedstocoarserresolutions,butratherthan
startoverontheoriginalwaveform,thecomputationusesasmoothed
versionofthefineresolutionwaveform
Thissmoothedversionisobtainedwiththehelpofthescalingfunction
Actually,thescalingfunctionissometimesreferredtoasthesmoothing
function
ng
ad
e
Thedefinitionofthescalingfunctionusesadilationoratwo
scaledifferenceequation:
.G
ho
wherec(n) isaseriesofscalarsthatdefinesthespecificscaling
function
r.R
.B
IntheDWT,thewaveletitselfcanbedefinedfromthescaling
function:
whered(n) isaseriesofscalarsthatarerelatedtothewaveformx(t)
FilterBanks
r.R
.B
.G
ho
ng
ad
e
Formostsignalandimageprocessingapplications,DWT
basedanalysisisbestdescribedintermsoffilterbanks
Theuseofagroupoffilterstodivideupasignalintovarious
spectralcomponentsistermedsubband coding.
ThemostbasicimplementationoftheDWTusesonlytwo
filters
r.R
.B
.G
ho
ng
ad
e
Thewaveformunderanalysisisdividedintotwocomponents,ylp(n)
andyhp(n),bythedigitalfiltersH0()andH1()
Thespectralcharacteristicsofthetwofiltersmustbecarefully
chosenwithH0() havingalowpass spectralcharacteristicand
H1()ahighpass spectralcharacteristic
Thehighpass filterisanalogoustotheapplicationofthewaveletto
theoriginalsignal,whilethelowpass filterisanalogoustothe
applicationofthescalingorsmoothingfunction
Theoriginalsignalcanoftenberecovered,butbothsubband signals
willrequired
Asecondpairoffilters,G0()andG1(),operateonthehighand
lowpass subband signalsandtheirsumisusedtoreconstructa
closeapproximationoftheoriginalsignal,x(t)
TheFilterBankthatdecomposestheoriginalsignalisusually
termedtheanalysisfilterswhilethefilterbankthatreconstructs
thesignalistermedthesynthesesfilters
FIRfiltersareusedthroughoutbecausetheyareinherentlystable
andeasiertoimplement
ad
e
ng
ho
.G
r.R
.B
Iftheoutputisessentiallythesame,asoccursinsomedatacompression
applications,theprocessistermedlossless,otherwiseitisalossy operation
Problemisthatdatasamplesgetdoubledhenceweusedownsampling
illustratedschema callybythesymbol2
Ifdownsampling isused,thentheremustbesomemethodforrecoveringthe
missingdatasamples(thosewithoddindices)inordertoreconstructthe
originalsignal
Anoperationtermedupsampling (indicatedbythesymbol2)accomplishes
thisoperationbyreplacingthemissingpointswithzeros
ad
e
ng
ho
.G
r.R
.B
ng
ho
.G
r.R
.B
Formostofthesignalanalyses,
theDWToperationtakesthe
formoflogarithmictree
Thebandwidthofthesignalis
halvedaftereachlevelof
decompositionalsoitismore
appropriatetodescribethe
frequencyinradiansinthe
discretedomain
Effectivelytheresolutionofthe
signal,whichistheamountof
detailinformationinthesignal,
ischangedbythefiltering
operationsandthescaleis
increasedbydownsampling
operations
ad
e
SignalDecomposition
Denoising
r.R
.B
.G
ho
ng
ad
e
Processingisdoneonthesubband signalsbefore
reconstruction
Thebasicassumptioninthisapplicationisthatthenoiseis
codedintosmallfluctuationsinthehigherresolution(i.e.,
moredetailed)highpass subbands
Thisnoisecanbeselectivelyreducedbyeliminatingthe
smallersamplevaluesinthehigherresolutionhighpass
subbands
Thetwohighestresolutionhighpass subbands are
examinedanddatapointsbelowsomethresholdare
zeroedout
Thethresholdissettobeequaltothevariance ofthe
highpass subbands
ad
e
ng
ho
.G
r.R
.B