Automatic Artifact Reduction Based on MEMD- for Seizure Prediction

Abstract

The performance of seizure prediction is usually affected by various kinds of artifacts, especially by physiological artifacts. To improve the performance of seizure prediction, this paper proposed an automatic artifact reduction method based on multivariate empirical mode decomposition and independent component analysis (MEMD-ICA). The proposed method could identify electrooculography (EOG) and electromyographic (EMG) artifacts precisely while keeping the useful neural signals as much as possible. The performance of seizure prediction has been significantly improved with an accuracy of 90.59% and a sensitivity of 91.09% based on CHB-MIT database.