Estimation and localization of multiple dipole sources for noninvasive mapping of muscle activity

In this paper, we develop a method for estimating and localizing multiple current-dipole sources within a finite-element arm model from surface EMG measurements. This extends previous work (Jesinger and Stonick, 1994 and LoPresti et al., 1995) in which we developed a realistic finite-element arm model from magnetic resonance images and then solved the single source localization problem via exhaustive search. Exhaustive search, however, is unacceptable for the multi-source localization problem due to the combinatorial nature of the number of solutions. (Our finite-element arm model has on the order of 10¹³ possible solutions.) Thus, we present an iterative technique which alternately pares down the size of the solution space and then locates sources one at a time. Examples demonstrating the power of our algorithm are included