SPATIOTEMPORAL MODELING OF NEUROMAGNETIC DATA .2. MULTISOURCE RESOLVABILITY OF A MUSIC-BASED LOCATION ESTIMATOR

A MUltiple SIgnal Classification-based (MUSIC) approach for neuromagne tic multi-source localization (Mosher ct al. [1992] (IEEE Trans Med En g BME-39:541-557) was evaluated through numerical simulations and by a pplying it to visually evoked neuromagnetic responses. A series of two -dipole and three-dipole spatio-temporal data sz ere generated to exam ine effects of 1) source configurations, 2) temporal correlations, 3) noise, and 4) subspace dimensionality assumptions on the number of MUS IC metric maxima, their amplitudes, and how the resulting metric maxim a locations relate to the actual source locations. Ln its present form , i.e., using simple one-dipole scanning over an assumed source subspa ce, MUSIC resulted either in 1) peaks sufficiently close to 1, but few er than the actual number of sources which affected location estimatio n accuracy, or 2) the peaks were too low to qualify as source location s. Our simulations indicate difficulties in defining threshold values as to which peak values are close enough to 1 while avoiding significa nt type IS errors (i.e., accepting peaks which should not be interpret ed as source locations). Modifications to the MUSIC approach are neces sary in order for the method to be considered of practical value for r eliably localizing multiple neuromagnetic sources in empirical cases i n which a high degree of temporal correlation between sources is likel y (e.g., visual data). (C) 1997 Wiley-Liss, Inc.