Epileptic Seizure Detection Based on ECoG Signal.

The article presents a system for detection of epileptic seizures with the use of automatic ECoG signal analysis. Features in the form of signal energy in nine different signal subbands were used. Next algorithms of feature selection (SFS, SBS) and classification (LDA, QDA) were performed. The developed algorithm was tested on signals taken from the ECoG database, containing signals recorded during epileptic seizures and without seizures. The proposed method is characterized by relatively low computational complexity and satisfactory accuracy of classification - up to 98%.