Key result
A novel seizure prediction method using sub-wave reconstruction and mRMR feature selection showed better prediction accuracy on the CHB-MIT EEG database compared with several relevant researches.
Why the study?
Seizure prediction from EEG is challenging due to non-linear and non-stationary signals, sudden attacks, and varying pathogeneses across patients.
Does the proposed method using sub-wave reconstruction and mRMR feature selection improve seizure prediction accuracy compared to existing methods?
Does the proposed method using sub-wave reconstruction and mRMR feature selection improve seizure prediction accuracy compared to existing methods?
A novel EEG-based seizure prediction method using wavelet transform and mRMR feature selection demonstrates improved accuracy over existing methods.
May aid seizure prediction algorithms; hypothesis-generating and requires prospective validation before clinical use.
Epilepsy is a group of neurological disorders characterized by recurrent epileptic seizures. In epilepsy monitoring, efficient seizure prediction based on the electroencephalogram (EEG) is expected to forecast epileptic seizures before they occur to protect the patient from further injuries. However, this problem is challenging as EEG signals are non-linear and non-stationary, seizures are sudden attacks, and the pathogeneses of different patients are different. Considering that some specific sub-waves, such as spike-wave, sharp-wave, may have gradually changed before a seizure takes place, this paper proposes a novel method to recognize pre-ictal and interictal signals from EEG for the prediction. Wavelet transform is adopted to reconstruct several sub-waves targeting spike-wave, sharp-wave, etc., from the original EEG. The maximum relevance and minimum redundancy (mRMR) principle is applied to find an optimal feature set for the subsequent classification. Then the selected multiple features from the sub-waves and those from the original signals are used as input of the neural network classifier to recognize the pre-ictal period for seizure prediction. We evaluated the proposed approach on the CHB-MIT EEG database. The proposed method shows better prediction performance in accuracy compared with several relevant researches.
No takes yet. Share an insight, caveat, or question.
Yan et al. (2021) studied Epilepsy. Seizure prediction based on specific sub-waves reconstruction and maximum-relevance, minimum-redundancy (mRMR) feature selection vs. Other relevant research methods was evaluated on Prediction performance in accuracy. A novel seizure prediction method using sub-wave reconstruction and mRMR feature selection showed better prediction accuracy on the CHB-MIT EEG database compared with several relevant researches.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: