Evaluation of Temporal and Spatial EEG Spike Detection Algorithms

The EEG (electroencephalogram) is an important clinical aid for the diagnosis of epilepsy, since it can reveal typical epileptiform activity. Since the visual identification of this activity is very time consuming, the automatic detection of this activity would be highly appreciated. However, since no universal EEG data set exists to evaluate a detection method, and because the performance of a method is highly dependent on the particular data set used, no comparison between the different methods published in the literature is possible. This study evaluates a number of different EEG spike detection algo- rithms (temporal as well as spatial methods) on the same data set. We con- clude that: (i) a simple basic temporal approach gives the highest detection performance, and (ii) spatial methods as such do not result in a gain of de- tection performance; an improvement may only be expected by including temporal information in these methods. Keywords—Spike detection, EEG