Bioelectric sources estimation using spatio-temporal matching pursuit

The bioelectric inverse problem is ill-posed and cannot be solved without physiological constraints on the spatio-temporal characteristics of the generators' activity. Decomposing the surface-recorded signal into their generators' temporal activity pattern is an important step towards a physiologically feasible solution of this problem. The new hybrid spatio-temporal matching pursuit (SToMP) algorithm for multiple source estimation of bioelectrical activity includes iterations of the following two stages. At the first stage of each iteration the multichannel signals are decomposed into the best-matched spatio-temporal waveforms selected from a physiologically motivated time-frequency dictionary. This spatio-temporal decomposition enables a fully linear exhaustive search for the optimal sources of each wave-form in the second stage of the algorithm. The linear exhaustive search is constrained to a three-dimensional non-uniform grid (or voxels) of all the anatomical candidates for sources. The SToMP algorithm was evaluated by simulation and was applied for source localisation of visual evoked potentials with MRI data constraints and visualisation.