Item

Abstract

The aim of the study is to apply spectral methods to functional magnetic resonance imaging (fMRI) data to reveal network structures and their temporal dynamics in the human brain. Several studies show that spectral methods are appropriate for investigating the temporal behavior of the BOLD signal ( M ARCHINI and R IPLEY 2000, M ÜLLER et al. 2001). BOLD stands for Blood Oxygenation-Level Dependent. In contrast to correlation methods in the temporal domain, spectral methods provide measures that are independent of the phase. Using the spectral density matrix, the sample coherence of time series can be estimated. The coherence provides a measure of the degree of linear association of time series. Moreover, the phase shift can be computed separately. The resulting maps of coherence and phase give detailed information on brain regions that belong to a structure with a similar vascular network and show the temporal behavior of the BOLD response function. The paper discusses the results of the method comparing data using different lengths of visual hemifield stimula - tion. Because the reason for the different temporal behaviour of the BOLD response in different brain regions is still unknown, the relation between phase shifts and signal amplitudes was studied. It could be revealed that the height of the signal change depends on the phase shift influenced by the length of the stimulation.