|
Abstract:
|
By generating high quality "movies" of the brain in action, functional Magnetic Resonance Imaging (fMRI) helps us to determine which parts of the human brains are activated by different task performances. Independent Component Analysis (ICA) has been successfully applied in analyzing fMRI data, to recover all the different source signals from different parts of the brain. However, due to the high sensitivity of the MR scanner, outliers are hardly evitable in acquiring the fMRI datasets, while they cause misleading effects for the analysis. In this paper, we introduce a robust ICA procedure, which uses robust singular value decomposition (rSVD) as a data-reduction step before applying ICA. We illustrate this method by both simulated data and a real fMRI dataset. Our method is proven to be powerful and advantageous for handling the outlier-situation.
|
