Modeling the dynamics of human neuromagnetic brain rhythms

No Thumbnail Available

URL

Journal Title

Journal ISSN

Volume Title

Helsinki University of Technology | Diplomityö
Checking the digitized thesis and permission for publishing
Instructions for the author

Date

2008

Major/Subject

Informaatiotekniikka

Mcode

T-61

Degree programme

Language

en

Pages

54 s.

Series

Abstract

As a precursor towards modelling event-related modulation of brain rhythms obtained by magneto encephalography (MEG), an oscillatory response function (ORF) is proposed. Analogous to the hemodynamic response function (HRF) in functional magnetic resonance imaging (fMRI), the ORF is a transformation applied to a boxcar function representing the stimulus time course, to predict the modulation of rhythmic brain activity. Both, linear and nonlinear parametric models for the ORF were derived in a generalized convolution framework. The corresponding convolution kernels were expanded as bilinear combinations of an orthonormal basis of discrete-time Laguerre functions. To estimate the model parameters, MEG data were acquired from 10 subjects during bilateral pneumotactile stimulation at two different stimulus rates in blocks of four different durations. The envelope of rhythmic activity in the 17-23 Hz frequency band was computed using the Hilbert transform, and subsequently averaged across blocks of each stimulus rate and duration. From a single representative channel over the rolandic region in each hemisphere, subject wise predictive models of envelope dynamics were derived. To study the generalizability of these models, MEG data were recorded from 5 different subjects with a different bilateral pneumotactile stimulation paradigm. A boxcar function was compared with the ORF-transformed boxcars as predictors of cortical minimum-norm current envelopes in a general linear model. The ORF-transformed boxcars localized rolandic activation in the 17-23 Hz band for tactile stimuli, to the primary motor cortex, better than the boxcar itself. As the models were able to predict well-known cortical generators of event-related mu rhythms, it is a worthwhile exercise to model the dynamics of rhythmic brain activity.

Description

Supervisor

Kaski, Samuel

Thesis advisor

Hari, Riitta
Parkkonen, Lauri

Keywords

event-related dynamics, generalized convolution, magnetoencephalography, mu rhythms, oscillatory response, predictive models, Volterra kernels

Other note

Citation