Variation-based Sparse Cortical Current Density imaging in estimating cortical sources with MEG data

Abstract

We investigated the performance of a new sparse neuroimaging method, i.e., Variation-Based Sparse Cortical Current Density (VB-SCCD) using magnetoencephalography (MEG) data to reconstruct extended cortical sources and their spatial distributions on the cortical surface. We conducted Monte Carlo simulation studies to compare the performance of the VB-SCCD method with different number of cortical sources and different number of MEG sensors. Our simulation data suggests that the VB-SCCD method is able to reconstruct extended cortical sources with the overall accuracy, while it has significantly reduced performance when cortical sources are radially oriented to MEG sensors. It has higher accuracy when the number of sensors is large and the source configuration is simple. We further assess the performance of VB-SCCD in real MEG data from an epilepsy patient and reconstructed cortical sources behind interictal spikes from the patient which are consistent with the clinical evaluation outcomes. This data indicate its promising applications in clinical problems related to neurological disorders.