On methods for pattern recognition with application to epileptic electroencephalograms

Patrik Wahlberg

On methods for pattern recognition with application to epileptic electroencephalograms

Patrik Wahlberg

Institutionen för elektro- och informationsteknik

Forskningsoutput: Avhandling › Doktorsavhandling (sammanläggning)

Sammanfattning

Popular Abstract in Swedish
Elektroencefalogrammet (EEG), dvs mätning av potentialer på skalpen, är en relativt enkel och gammal teknik för att studera hjärnans aktivitet. Trots de stora framstegen inom s k brain-mapping-tekniker på senare år (magnetröntgen, PET, SPECT), är EEG fortfarande ett viktigt verktyg för diagnos. En av dess fördelar gentemot andra tekniker är dess fina tidsupplösning.

Ett stort användningsområde för EEG-signalen är diagnosticering av epilepsi. Epilepsi yttrar sig i EEG på framför allt två olika sätt: vid anfallsstart sker en förhöjning av signalenergin, och mellan anfallen kan förekomma spontana transienter, s k spikar. Spikarna är viktiga eftersom man antar att de ofta härrör från epileptiska fokus. Även anfallsstarter används för lokalisation av fokus.

Denna avhandling behandlar olika tekniker för signalbehandling av flerkanaligt uppmätta EEG-signaler från epileptiker, närmare bestämt spikar och anfallsstarter. För att få kunskap om de epileptiska fokusen, t ex deras antal och placering i hjärnan vill man kategorisera spikar. Eftersom kategorierna ofta är illa definierade då de uppmäts på skalpen krävs statistisk robustitet hos kategoriseringsalgoritmen. Dessutom önskar man att spikar av samma kategori är synkroniserade till varandra innan man utför kategorisering, eftersom annars kategorierna kan vara svåra att finna. Den rika förekomsten av störning i signalerna påkallar metoder för brusreduktion. Dessa avgränsade problem studeras i tre av avhandlingens fem delar. Den sista delen handlar om brusreduktion i anfallsstart-EEG, vilket kräver tekniker avpassade till icke-stationära signaler eftersom anfallsstart-EEG har ett snabbt varierande frekvensinnehåll.

Originalspråk	engelska
Kvalifikation	Doktor
Tilldelande institution	Institutionen för elektro- och informationsteknik
Handledare	[unknown], [unknown], handledare, Extern person
Tilldelningsdatum	1999 okt. 15
Förlag	Patrik Wahlberg, Department of Applied Electronics, Lund University, P.O. Box 118, S-221 00 Lund, Sweden,
Status	Published - 1999

Bibliografisk information

Defence details

Date: 1999-10-15
Time: 13:15
Place: Room E:1406, E-huset, LTH, Lund

External reviewer(s)

Name: Koski, Timo
Title: Docent
Affiliation: [unknown]

---

Ämnesklassifikation (UKÄ)

Elektroteknik och elektronik

Citera det här

@phdthesis{bd672dd8896b4c82b069cd2c3abddb2c,

title = "On methods for pattern recognition with application to epileptic electroencephalograms",

abstract = "The thesis treats methods for pattern recognition in multichannel electroencephalogram (EEG) signals, for application to diagnostics of epilepsy. Parts I-IV treat methods for feature extraction and clustering of EEG spikes, occurring between epileptic seizures, and part V presents a method for filtering seizure onset EEG signals. In part I, Hermite functions are used for parametric description of multichannel spikes, and the L-means method for clustering. The method is evaluated on real life signal sets. Part II treats the problem of making clustering methods statistically robust, and evaluates the fuzzy L-means method and the k-component graph-theoretic method in this respect. The effect of incorporation of the physical geometry of the electrode positions is also studied. In part III, the problem of alignment of the signals that are to be clustered is treated. Two new estimators are suggested and evaluated on simulated as well as real life signals. Part IV treats denoising of signal matrices under the assumption that the signals have low rank and the noise has full rank. A method is suggested consisting of representation in a basis of unit rank matrices, and algorithms for determination of such bases are given. Applicability to real life signals is demonstrated. Part V describes a technique for approximate Wiener filtering of non-stationary seizure onset EEG signals. The method relies on a time-frequency domain identity which is approximately valid for underspread stochastic processes, and estimates the Weyl symbol of the Wiener filter with aid of coherence functions between channels.",

keywords = "clustering, EEG signal processing, epilepsy, statistical robustness, feature extraction, signal alignment, denoising, Wiener filtering, time-frequency analysis., Signal processing, Signalbehandling",

author = "Patrik Wahlberg",

note = "Defence details Date: 1999-10-15 Time: 13:15 Place: Room E:1406, E-huset, LTH, Lund External reviewer(s) Name: Koski, Timo Title: Docent Affiliation: [unknown] ---",

year = "1999",

language = "English",

publisher = "Patrik Wahlberg, Department of Applied Electronics, Lund University, P.O. Box 118, S-221 00 Lund, Sweden,",

school = "Department of Electrical and Information Technology",

}

TY - THES

T1 - On methods for pattern recognition with application to epileptic electroencephalograms

AU - Wahlberg, Patrik

N1 - Defence details Date: 1999-10-15 Time: 13:15 Place: Room E:1406, E-huset, LTH, Lund External reviewer(s) Name: Koski, Timo Title: Docent Affiliation: [unknown] ---

PY - 1999

Y1 - 1999

N2 - The thesis treats methods for pattern recognition in multichannel electroencephalogram (EEG) signals, for application to diagnostics of epilepsy. Parts I-IV treat methods for feature extraction and clustering of EEG spikes, occurring between epileptic seizures, and part V presents a method for filtering seizure onset EEG signals. In part I, Hermite functions are used for parametric description of multichannel spikes, and the L-means method for clustering. The method is evaluated on real life signal sets. Part II treats the problem of making clustering methods statistically robust, and evaluates the fuzzy L-means method and the k-component graph-theoretic method in this respect. The effect of incorporation of the physical geometry of the electrode positions is also studied. In part III, the problem of alignment of the signals that are to be clustered is treated. Two new estimators are suggested and evaluated on simulated as well as real life signals. Part IV treats denoising of signal matrices under the assumption that the signals have low rank and the noise has full rank. A method is suggested consisting of representation in a basis of unit rank matrices, and algorithms for determination of such bases are given. Applicability to real life signals is demonstrated. Part V describes a technique for approximate Wiener filtering of non-stationary seizure onset EEG signals. The method relies on a time-frequency domain identity which is approximately valid for underspread stochastic processes, and estimates the Weyl symbol of the Wiener filter with aid of coherence functions between channels.

AB - The thesis treats methods for pattern recognition in multichannel electroencephalogram (EEG) signals, for application to diagnostics of epilepsy. Parts I-IV treat methods for feature extraction and clustering of EEG spikes, occurring between epileptic seizures, and part V presents a method for filtering seizure onset EEG signals. In part I, Hermite functions are used for parametric description of multichannel spikes, and the L-means method for clustering. The method is evaluated on real life signal sets. Part II treats the problem of making clustering methods statistically robust, and evaluates the fuzzy L-means method and the k-component graph-theoretic method in this respect. The effect of incorporation of the physical geometry of the electrode positions is also studied. In part III, the problem of alignment of the signals that are to be clustered is treated. Two new estimators are suggested and evaluated on simulated as well as real life signals. Part IV treats denoising of signal matrices under the assumption that the signals have low rank and the noise has full rank. A method is suggested consisting of representation in a basis of unit rank matrices, and algorithms for determination of such bases are given. Applicability to real life signals is demonstrated. Part V describes a technique for approximate Wiener filtering of non-stationary seizure onset EEG signals. The method relies on a time-frequency domain identity which is approximately valid for underspread stochastic processes, and estimates the Weyl symbol of the Wiener filter with aid of coherence functions between channels.

KW - clustering

KW - EEG signal processing

KW - epilepsy

KW - statistical robustness

KW - feature extraction

KW - signal alignment

KW - denoising

KW - Wiener filtering

KW - time-frequency analysis.

KW - Signal processing

KW - Signalbehandling

M3 - Doctoral Thesis (compilation)

PB - Patrik Wahlberg, Department of Applied Electronics, Lund University, P.O. Box 118, S-221 00 Lund, Sweden,

ER -

On methods for pattern recognition with application to epileptic electroencephalograms

Sammanfattning

Bibliografisk information

Ämnesklassifikation (UKÄ)

Fingeravtryck

Citera det här