Author:

Keywords:

Magnetic resonances, Nervous system. Sensory organs

Abstract:

This thesis discusses two new contributions to the analysis of fMRI (functional Magnetic Resonance Imaging) signals. The signals were obtained by scanning the brains of primates using an MRI scanner. The first contribution is an enhanced cluster tool. K-means and Fuzzy Cluster Analysis are often used for fMRI signal analysis to detect activity patterns in the brain, but they are based on linear concepts. Conversely, our novel cluster tool is based on the Procaccia-Grassberger theorem, which is well-known in the field of non-linear state dynamics. The obtained results show that our method is better adapted to handle the non-linear nature of fMRI signals, and that it is less sensitive to noise, which makes that the obtained clusters are more related to brain regions that are responding differentially to the applied stimuli. The second contribution is an extension to nonparametric statistical tests such as the Mann-Whitney test, the Kolmogorov-Smirnov test (KS) and the Cramèr-von Mises test (CvM). Statistical tests are used to check whether a given brain region is significantly responsive to a (set of) stimuli. Our extension corrects for the omni-present autocorrelations in fMRI signals, which consequently enables the use of these nonparametric tests for fMRI signals analysis. With respect to the current, parametric tests like the Student t-test, which can only detect a difference in the average activation, the KS and CvM tests can detect differences in the nature of the activation. In combination with the traditionally-used statistical tests, this is a source of additional information. Together with the novel cluster tool we developed, this contribution is expected to increase our insight into the functioning of the brain.