A New Approach for Separation of Functional MRI Images´ Components Using Modified FastICA Algorithm

This paper proposes a new approach for separation of components in functional MRI sequences using FastICA. In this paper we will demonstrate that if we subtract background (which is separated from sources by PCA) from other principal components, the algorithm converges very fast. The proposed method is more robust and much more computationally efficient than methods, which previously has been applied for separation of components in functional MRI sequences.