An Iterative ICA Method for Separating fMRI Acoustic Noise Components

A new iterative algorithm for separating mixtures of multi-channel signals is proposed. This algorithm extends the instantaneous independent component analysis algorithm to multi-channel blind source separation algorithm. Separation is processed by decomposing convolutive mixtures to instantaneous mixtures. Simulation results for real fMRI (functional magnetic resonance imaging) scanner noise show that the proposed algorithm is very effective in blind source separation.