Seizure detection using empirical mode decomposition and time-frequency energy concentration

Abstract

The aim of this study is to evaluate a new method for seizure detection using the tripolar Laplacian electroence-phalography signal (tEEG) recorded using a tripolar concentric ring electrode (TCRE) on the scalp surface of rats based on empirical mode decomposition (EMD) and time-frequency energy concentration. Data from 10 rats were examined with the proposed algorithm. After EMD decomposition, three oscillation components named intrinsic mode functions (IMFs) were selected. An energy estimate of the TFR for the selected IMFs was calculated and used as a feature for automatic seizure detection of the tEEG signals. After classification the obtained results using the proposed method produced an accuracy of 98.61%. This study developed the proposed algorithm to work with TCREs, and shows it to be effective to detect seizures from rat's tEEG signals.