Computer-aided analysis of the epileptic EEG

An introductory review of a comprehensive approach to the problem of automatic evaluation of the epileptic EEG in the frame of syntactic analysis is presented. The aims and merits of the syntactic approach are discussed and the viability of the method is demonstrated on a few chosen examples. Working procedures for automatic segmentation and classification of patterns in single- as well as in multi-channel case of the epileptic EEG are briefly described. These include algorithms for segmentation by a finite-state automaton technique (single channel) and, in multi-channel case, a sparsely updated Kalman filter algorithm. Inherent possibilities of the approach presented are shown to open the way for a global description of a seizure in terms of the Markov chains theory.