SIMULTANEOUS SPIKE DETECTION AND TOPOGRAPHIC CLASSIFICATION IN PEDIATRIC SURFACE EEGS

IN this study, an algorithm is introduced for the automatic detection and simultaneous topographic classification of interictal regional spi ke activity in pediatric surface EEG records. The algorithm is based o n the classification of the topographic distribution of instantaneous power by means of a 'group' trained classifier. The results of automat ic spike with the decisions of two experienced electroencephalographer s. Four routine EEG records exhibiting (multi)regional spikes were exa mined. The mean selectivity for the automatic spike detector was 84.6% (mean sensitivity 88.1%, mean specificity 89.3%) and for the electroe ncephalographers 85.3%. All spikes detected by the algorithm were simu ltaneously classified according to their topographic characteristics. The results of automatic spike classification (lateralization/localiza tion) corresponded to the results of visual analysis.