A test of brain electrical source analysis (BESA): a simulation study

Abstract

The present report summarizes the results of a simulation study on the accuracy of Scherg's implementation of spatio-temporal analysis (BESA) for estimating the parameters (wave shape, location, orientation) of the intracranial sources of event-related brain potentials (ERPs) recorded from the scalp. In view of the subjective factors that might influence a solution, 10 subjects, ranging from those with much experience with ERPs and extensive background in the use of BESA to those with little experience with BESA and/or no knowledge of ERPs, independently analyzed a set of simulated somatosensory ERP data. The simulation contained wave forms from 32 electrode sites generated by a combination of 10 dipole sources. The primary question was how faithfully the different subjects would depict the source wave shapes, locations and orientations. Based on the 9 subjects who were familiar with ERPs, the grand-average cross-correlation coefficient between subjects' estimated and actual source wave shapes was 0.89 (standard deviation (S.D.)=0.17). The grand-average location error, based upon a head diameter of 17 cm, was 1.4 cm (S.D. = 1.0 cm). The grand-average orientation error was 24.4° (S.D.=20°).