Epileptic MEG spikes detection using common spatial patterns and linear discriminant analysis

Abstract

Epilepsy is a brain disorder, where patients' lives are extremely disturbed by the occurrence of sudden unpredictable seizures. This paper develops patient-independent signal processing techniques based on common spatial patterns and linear discriminant analysis to detect epileptic activities (spikes) from multi-channel brain signal recordings. In contrast to current existing studies which heavily rely on the analysis of electroencephalogram (EEG) data for the detection of epileptic activities, this research work considers magnetoencephalography (MEG) recordings for the detection of epileptic spikes. This requires careful development since unlike EEG spikes, MEG spikes do not have well-defined morphological characteristics. Due to the recent advances in MEG technology, it became possible to consider MEG signals to detect and analyze epileptic activities, but efforts to develop signal processing tools in this area are still in its outset, as compared with those devoted to EEG signal processing.