Accurate localization of electrical sources in the brain using EEC's or MEG's measured on the surface of the head has many important clinical and research applications. Localization is most simply and easily done by calculating an inverse solution for a source in a spherical model of the head which produces EEC's or MEG's that most closely match the measured values. The location of the solution is taken to indicate the location of the actual source. However, modeling errors caused by the differences between the geometry of actual heads and a spherical model produce localization errors. The first specific aim of this research project is to determine the effects of such model in errors on localization accuracy and to determine the optimal conditions for using a spherical head model, i.e., the conditions that produce the best localization accuracy. Investigations will be performed to determine the optimal model parameters, measurement grid size and location, measurement point density, etc. Recently, methods have been developed as part of this research project which make it practical to calculate inverse solutions in realistic head models. This presents the opportunity for significant improvements in localization accuracy. The second specific aim of this project is to determine the improvement in localization accuracy that can be achieved by using realistic models and, as for the spherical model studies above, determine optimal conditions for using realistic models. The realistic models will be generated from X-rays, MRI's, CT's, etc. of actual subjects. The third specific aim of this project is to develop guidelines for the selection of the appropriate head model to be used for various localization tasks. These guidelines will indicate when a simple spherical head model is adequate or when a realistic model must be used. In addition, these guidelines will indicate the conditions that must be met in order to achieve improved accuracy using a realistic model. These conditions would include such fairs as adequate number of EEC or MEG measurements, adequate signal-to-noise ratio of the measurements, etc.
