Brain electrical tomography using algebraic reconstruction techniques and Tikhonov regularization

Abstract

The inverse EEG problem is solved using simulated potentials, through an analytic technique providing information about extended intracranial distributions, with separate source and sink positions. A three-layered concentric sphere model is used for representing head geometry. Comparative performance evaluation of the Algebraic Reconstruction Techniques (ART) and the Tikhonov Regularization Technique (TRT) is performed. ART algorithms specifically designed to compensate for noisy data perform similarly with TRT, but require the prior knowledge of the characteristic of the noise affecting the data. The empirical composite residual and smoothing operator (CRESO) criterion provides an approximation to the optimal regularization parameter t of the TRT, without requiring any prior knowledge about the noise in measured potentials. Therefore, when the CRESO criterion is successful in providing a t value. TRT may be used in real EEG data inversions for the creation of brain electrical activity tomographic images.