This phase II proposal will continue the implementation of Lead Field Analysis (LFA) to compute dipole source localization (DSL) within realistic head models. In Phase I the superiority of LFA over conventional boundary element computations was demonstrated in terms of computational speed and storage requirements. This methodological advance, when coupled with the possibility of decomposing the DSL algorithm into (1) a computationally intensive portion that characterizes the field potentials and current fluxes on the inner shell of a realistic model advantage and (2) the less computationally demanding estimation of weighing coefficients dependent upon the location and orientation parameters of each dipole, make it possible to implement DSL using the boundary element method (BEM) with realistic head models.
The Specific Aims of Phase II are (1) To test the robustness of the LFA method when used with realistic head shapes, so that it can be used as a module in existing DSL programs. These tests will include studying the accuracy and performance of BEM solver over a wide range of modelling parameters of the human head. (2) To develop a central computational unit, based on a high performance workstation, to compute potentials and normal currents on the inner surface of the skull from model data supplied by users. (3) To develop a software module which will run on a PC and compute the lead fields at any desired point in the brain based upon the inner-shell currents and potentials obtained from the central unit. This module will be designed as a subroutine which will be callable from any DSL program or incorporated into DSL software from Abratech.
