Application of independent component analysis for activation detection in functional magnetic resonance imaging (FMRI) data

In this extended summary, our aim is analyzing functional magnetic resonance imaging (fMRI) data by independent component analysis (ICA) in order to find regions of brain which were activated by neural activity in human brain. We employ the minimum description length (MDL) criterion to reduce the dimension of the data and estimate the number of components, which makes ICA work more efficiently. We also use a simple oscillating index method to select automatically the components of interest. MDL and oscillating index criteria have not already been used in applying ICA for analyzing fMRI data. In order to investigate the advantage of using MDL and oscillating index, we perform some experiments for both simulated and experimental fMRI dataset and show the results. In order to justify the performance, receiver operating characteristic (ROC) curve have been drawn.