fMRI and MRI Image Registration and Statistical Mapping

Registration is required to integrate the information gained from different image modalities acquired from the same patient to obtain a comprehensive interpretation. In fMRI the acquired brain volume data is often distorted by head motion. As a result accurate information regarding task-induced (or resting-state) signal changes cannot be extracted. So registration is essential among fMRI volumes (intra modal registration) to correct for such motion. Moreover to detect the active voxels of the brain due to a particular task multimodal registration between fMRI and MRI is required. This thesis work is divided into two parts: 1) The first part presents the application of an automatic mutual information based image registration technique to correct for the motion in head brain volumes acquired by fMRI and also to register those volumes to brain volumes acquired by MRI. 2) The second part presents the application of statistical mapping based on the General Linear Model (GLM) to detect functional activities in the brain due to task related stimulation. Finally three sets of fMRI data have been analyzed and the results are presented.