RPCA-Based Noise Suppression in MEG Measurement for Improving Bio-Electromagnetic Source Estimation

Magnetoencephalography (MEG) is a promising technology, which could be used in a variety of biomedical applications. However, MEG electromagnetic measurement is usually degraded by noise. Noise suppression in MEG measurement is particularly challenging because it is difficult to remove the noise and preserve the information components in the MEG data. In this study, a novel noise suppression method, based on robust principal component analysis (RPCA) technique, is presented and applied to the estimation of bio-electromagnetic field in source space for the first time. The proposed method gives a constrained optimization of MEG electromagnetic domain transformations such that the matrix of transformed MEG measurement can be decomposed as the sum of a sparse matrix of noise and a low-rank matrix of denoised data. Applying the proposed method to a number of simulations showed significant improvement of the result accuracy.