Blind Dereverberation for fMRI Noise Based on SIMO Linear Prediction Method

Room reverberation degrades performance of acoustic noise cancellation techniques for functional magnetic resonance imaging (fMRI) environment. In this paper, we present a method based on single input multiple output (SIMO) linear prediction to remove the reverberation from the observed broadband fMRI noise signal. A performance criteria is defined based on spectral characteristic of the signal and performance of the method is analyzed using two types of data, fMRI noise with and without speech signal.