Entwicklung, Testung und Anwendung von Verfahren zur nichtlinearen Analyse von kortikalen Aktivierungs- und Deaktivierungsmustern im EEG
Year: 2008
Type: dissertation
Abstract: The aim of this work was the development of two different time-variant \nnonlinear methods (1) to quantify the nonlinear stability and (2) to detect \nand quantify quadratic phase couplings (QPC). Starting point of the \nmethodical investigation of the nonlinear stability was the time-invariant \nestimation of the largest Lyapunov exponent, which was adapted on a \npointwise investigation. An optimization of critical parameter of this \nestimation was performed using simulations and real data. Nonlinear \nportions of the investigated data were detected by a test of nonlinearity \nusing a surrogate data approach. Starting point of the methodical \ninvestigation of QPC was the time-invariant, parametric estimation of the \nbispectrum. The time-variant approach was carried out by an estimation of \nAR parameters using the ”recursive instrumental variable” algorithm. The \noptimization of critical parameters of the estimation occured by \ndata-driven simulations. A two-dimensional approach of optimization was \nused. The aim of the practical investigation of the EEG (and ECoG, \nrespectively) was the combined application of both methods. Sleep pattern \nand the effect of vibroacustic stimulations (VAS) in the fetal ECoG, \nburst-interburst pattern (BIP) in the neonatal EEG and burst-suppression \npattern (BSP) in the EEG of sedated patients were examined. The analysis of \nthe fetal ECoG showed the occurrence of cyclic ECoG activity already at a \nvery young gestational age and contributed essentially to the understanding \nof the time course of the development of characteristic sleep states. VAS \nduring NREM sleep caused an arousal reaction in the fetal ECoG. The \ntime-course of this arousal could be quantified by the nonlinear stability. \nInvestigating the neonatal EEG during BIP, the existence of a 10-seconds \nrhythmicity in the time-course of QPC could be proven. High QPCs in the EEG \nof the sedated patients during BSP could be detected and quantified in its \ntime course. The combined application of both nonlinear methods was able to \ncontribute fundamentally to the amount of findings and could be utilized \ncomplementary in the interpretation of the results. \n \nkeywords: time-variant nonlinear stability, time-variant parametric \nbispectrum , fetal ECoG, Burst-Interburst-Pattern, \nBurst-Suppression-Pattern
Author: Karin Schwab
Institution: Technische Universität Ilmenau
Language: German
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2207947436
