RT Journal Article
SR Electronic
T1 Populational Super-Resolution Sparse M/EEG Sources and Connectivity Estimation
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 346569
DO 10.1101/346569
A1 E. Gonzalez-Moreira
A1 D. Paz-Linares
A1 E. Martinez-Montes
A1 P. Valdes-Hernandez
A1 Jorge Bosch-Bayard
A1 M.L. Bringas-Vega
A1 P. Valdés-Sosa
YR 2018
UL http://biorxiv.org/content/early/2018/06/13/346569.abstract
AB In this paper, we describe a novel methodology, BC-VARETA, for estimating the Inverse Solution (sources activity) and its Precision Matrix (connectivity parameters) in the frequency domain representation of Stationary Time Series. The aims of this method are three. First: Joint estimation of Source Activity and Connectivity as a frequency domain linear dynamical system identification approach. Second: Achieve super high resolution in the connectivity estimation through Sparse Hermitian Sources Graphical Model. Third: To be a populational approach, preventing the Inverse Solution and Connectivity statistical analysis across subjects as a postprocessing, by modeling population features of Source Activity and Connectivity. Our claims are supported by a wide simulation framework using realistic head models, realistic Sources Setup, and Inverse Crime effects evaluation. Also, a fair quantitative analysis is performed, based on a diversification of quality measures on which state of the art Inverse Solvers were tested.