The flicker electroretinogram in phase space [Elektronische Ressource] : embeddings and techniques / vorgelegt von Albrecht Johannes Rilk

Albrecht Johannes Rilk <[e-mail]>

Sujets

• Gesundheit

Informations

• Publié par: eberhard_karls_universitat_tubingen
• Publié le: 01 janvier 2003
• Nombre de lectures: 25
• Poids de l'ouvrage: 1 Mo

Extrait

AusderUniversitats Augenklinik¨ Tubingen¨
AbteilungAugenheilkundeII
¨ArztlicherDirektor: ProfessorDr.E.Zrenner
TheFlickerElectroretinogram
inPhaseSpace:
EmbeddingsandTechniques
Inaugural Dissertation
zurErlangungdesDoktorgrades
derMedizin
derMedizinischenFakultat¨
derEberhard Karls Universitat¨
zuTubingen¨
vorgelegtvon
ALBRECHT JOHANNES RILK
aus
Aalen
2003Dekan: ProfessorDr.C.D.Claussen
1.Berichterstatter: ProfessorDr. E.Zrenner
2. ProfessorDr. F.Nusslin¨Contents
TableofContents 6
Abbreviations 8
1 Introduction 8
1.1 Purposeandscope . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 ERGbasics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.1 NormalERG. . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.2 Clinicalapplications . . . . . . . . . . . . . . . . . . . . . . 12
1.2.3 ERGvarieties . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3 WhyyetanotherERGsystem? . . . . . . . . . . . . . . . . . . . . 14
1.4 Calypso . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4.1 Functionaloverview . . . . . . . . . . . . . . . . . . . . . . 16
2 Materialsandmethods 20
2.1 Thedelayembedding . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1.1 Constructingdelayvectors . . . . . . . . . . . . . . . . . . 21
2.1.2 Thelookofatrajectory . . . . . . . . . . . . . . . . . . . . . 23
2.1.3 Thetheoreticalpointofview . . . . . . . . . . . . . . . . . 24
2.2 Nonlinearprojectivenoisereduction . . . . . . . . . . . . . . . . . 26
2.2.1 NNRbasics . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.2 Estimatingtheparametervalues . . . . . . . . . . . . . . . 28
2.3 Reducingtherecordingtime . . . . . . . . . . . . . . . . . . . . . . 29
34 CONTENTS
2.4 Experimentalsetup . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1 Hardwareandsoftware . . . . . . . . . . . . . . . . . . . . 30
2.4.2 StandardERGrecordings . . . . . . . . . . . . . . . . . . . 31
2.4.3 Safetyconsiderations . . . . . . . . . . . . . . . . . . . . . . 31
2.5 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.5.1 Ethicalconsiderations . . . . . . . . . . . . . . . . . . . . . 32
2.6 Examinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7 Dataprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7.1 Frequencyfiltering . . . . . . . . . . . . . . . . . . . . . . . 34
2.7.2 Removingartifacts . . . . . . . . . . . . . . . . . . . . . . . 37
2.7.3 Averaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3 Results 40
3.1 StandardISCEVflickerERGembedding . . . . . . . . . . . . . . . 40
3.2 Flickertrajectorytopology . . . . . . . . . . . . . . . . . . . . . . . 42
3.2.1 Descriptiveapproach . . . . . . . . . . . . . . . . . . . . . . 42
3.2.2 Polarcoordinateshelpquantifying . . . . . . . . . . . . . . 44
3.3 PathologicalflickerERGs . . . . . . . . . . . . . . . . . . . . . . . 46
3.3.1 Usher’ssyndrome . . . . . . . . . . . . . . . . . . . . . . . 46
3.3.2 Retinitispigmentosa . . . . . . . . . . . . . . . . . . . . . . 47
3.3.3 Juvenilemaculardegeneration . . . . . . . . . . . . . . . . 49
3.3.4 Stargardt’sdisease . . . . . . . . . . . . . . . . . . . . . . . 50
3.4 NonlinearNoiseReductionperformance . . . . . . . . . . . . . . 51
3.4.1 Generatingthetestdatasets. . . . . . . . . . . . . . . . . . 52
3.4.2 Differentkindsofnoise . . . . . . . . . . . . . . . . . . . . 53
3.4.3 Iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.4.4 NNRonsinusoids . . . . . . . . . . . . . . . . . . . . . . . 55
4 Discussion 58
4.1 Topographicanglesynopsis . . . . . . . . . . . . . . . . . . . . . . 58CONTENTS 5
4.2 NNRandothernoisereductiontechniques . . . . . . . . . . . . . 60
4.2.1 NNRandaveraging . . . . . . . . . . . . . . . . . . . . . . 60
4.2.2 NNRandfrequencyfilters . . . . . . . . . . . . . . . . . . . 62
4.3 Synergies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.4 Whatcangowrong . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.4.1 Embeddingparameters . . . . . . . . . . . . . . . . . . . . 64
4.4.2 Artifacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5 Conclusion 70
5.1 Futureoutlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.1.1 Follow ups . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.1.2 Extendingthepolarcoordinateconcept . . . . . . . . . . . 72
5.1.3 Visualizingmulti dimensionalspaces . . . . . . . . . . . . 72
5.1.4 ThemultifocalERG . . . . . . . . . . . . . . . . . . . . . . . 74
A Noiseindifferentflavors 76
A.1 Signal independentnoisemodels . . . . . . . . . . . . . . . . . . . 77
A.1.1 Gaussiannoise . . . . . . . . . . . . . . . . . . . . . . . . . 77
A.2 Signal dependentnoisemodels . . . . . . . . . . . . . . . . . . . . 78
A.2.1 Spectrallymatchednoise . . . . . . . . . . . . . . . . . . . 79
B Standarddeviation,RMS,andSNR 81
C TheFourierTransform 83
ListofFigures 86
ListofTables 87
Bibliography 88
Danksagung 926 CONTENTS
Lebenslauf 93Abbreviations
ADC Analog to DigitalConverter
Calypso ComputerizedAnalysisofPhysiologicalSamplingsinOphthalmology
DSP DigitalSignalProcessor
ECG Electrocardiogram
ERG Electroretinogram
FFT FastFourierTransform
FIR FiniteImpulseResponse
IIR Infinite
ISCEV InternationalSocietyforClinicalElectrophysiologyofVision
mfERG MultifocalElectroretinogram
NNR NonlinearProjectiveNoiseReduction
PCA PrincipalComponentAnalysis
RMS RootMeanSquare
RP RetinitisPigmentosa =RetinopathiaPigmentosa
SMN SpectrallyMatchedNoise
SNR Signal to NoiseRatio
WGN WhiteGaussianNoise
7Chapter1
Introduction
bout one to three million people worldwide suffer from a form of
hereditary retinal degeneration, i.e., retinitis pigmentosa (RP); in Ger-Amany alone, 30,000 to 40,000 patients are affected. Presently, there is
stillnocureforRP,whichisthusoneofthemostfrequentcausesforblindness.
Only a few therapies are currently available that allow for partial inhibition of
retinal degeneration, but these are applicable to only some rare forms of RP.
However, the fact that some treatment does exist encourages us to search for
newwaysof,ifnotcuring,atleastinhibitingdegenerationinthemorefrequent
formsofRP.
thSince RP was first described in the 19 century, several different therapeutic
approaches have been attempted. In recent years, molecular genetic studies
have been able to shed some light on the pathophysiological disease mecha
nisms. Forexample,asofApril2003,thereare134mappedgenesand90cloned
genesthatarethoughttoberesponsibleforRP(DAIGER,2003). Inclinicalprac
tice, however, the only reasonable approach seems to qualify the pathological
changeswithintheretinaintermsofmorphology,and,ofcourse,torecordthesebyelectrophysiologicaltechniquesaspreciselyaspossible. Atreatment
is most effective when applied in the disease’s early stages, as is the chance of
preserving visual function. By means of objective electrophysiology, we can
detect pathological changes even before the patient himself becomes aware of
them. Butevensuchearlydetectionneedstobeimprovedupon.
Hence, the task of clinical research is to refine and to extend the electrophys
iological methodology with two aims in mind: On the one hand, we need to
81.1 PURPOSE AND SCOPE 9
preserve the patient’s life quality as much as possible by early diagnostics and
treatment; on the other hand, meaningful and diagnostically relevant results
are mandatory, especially in the disease’s final stages, when signals have be
come very weak or even seemingly extinguished. Achieving such aims would
be beneficialnot only forRP patients, but forall patients withheavily reduced
electroretinograms(ERG),e.g.,thosesufferingfromSTARGARDT’sdisease,and
fromotherretinaldystrophies,betheyhereditaryoracquired.
The diagnostic advantages by state of the art data processing available to oph
thalmology are far from being exhausted. A look at the literature reveals sev
eralmethodsthathaverecentlyemergedfromthenonlineardynamicssciences,
andthathavebeenshownusefulfor,e.g.,cardiologicalproblems. Butasofyet,
thesemethodsobviouslyhavenotbeenappliedtoophthalmologicaltasks.
1.1 Purposeandscope
In recent times, strong evidence was found that the representation of complex,
nonlinear data in a phase space embedding reveals more information than the
common amplitude time diagram. It has seemingly not yet been considered if
thisincludesERGdataaswell.
Therefore,thisthesiswillintroducesomenonlineartechniquestoERGanalysis,
and will, in particular, address the question of how ERG data may be embed
dedinamulti dimensionalspace;thisisbecausesuchanembeddingformsthe
startingpointforseveralmoderndataprocessingmethods. Moreover, theem
bedding itself can easily be visualized, showing ERG structures in a different
way. Therefore, such ERG embeddings will be explored and discussed, since
theymightaddtothepowerfuldiagnosticcapabilitiesanERGexaminationyet
provides.
Nonlinear analysis and chaos theory have been developed a bunch of tech
niques for cases where the examined system is not strictly deterministic, but
displays more complexity than can be handled by traditional methods. These