Independent Component Analysis of Functional Magnetic Resonance Imaging (fMRI) Data: A simple Approach

Independent Component Analysis (ICA) separates spatial and temporal components of fMRI data which may consist of activation patterns, cardiac and respiratory tasks and other artifacts. In this study sources of (fMRI) data are separated using ICA based on simple fixed point iteration method and steepest ascent method. Both are the simplest methods used in optimization. However in this study complete matrix W (un-mixing matrix) is updated in each iteration instead of vector based updating of W. This makes the source separation process very fast. Simulated fMRI data is processed using the proposed method and the results are compared with other ICA approaches in terms of speed and accuracy.