The Investigation of Channel Selection Effects on Epileptic Analysis of EEG Signals

A great number of methods are used in order to increase the speed of decision units in epileptic analysis of the multi-channel EEG signals. Channel selection is one of the main methods used for the reduction of the processing load. By eliminating the non-distinct channels, the performance of the system can be improved. In this study, the seizure detection performances of EEG signals obtained by 21 different channels were evaluated. This study was carried out patient-specifically for each six patients. The feature set is generated via calculating 26 features from EEG signals. The dimension of feature set for each channel is reduced using Principal Component Analysis. The reduced feature sets were divided as training and testing data using cross-validation method. With Linear Discriminant Analysis, the classification was done for each channel and performances of channel were compared. Depending on the channel selection, almost 9% differences in the classification accuracies have been observed.