An accurate epileptic seizure prediction algorithm can alleviate the problem and reduce risks in the life of a patient suffering from epilepsy. The main motive of this work is to propose a model which can predict seizures well in advance of its occurrence. Multivariate statistical process control (MSPC) has been used for seizure predictions in long-term scalp EEG signal. It has been observed that excessive neuronal activity in the preictal period of seizure changes the electrical characteristic from chaotic to rhythmic behavior. These changes have been utilized for prediction. Eight temporal based features are used for predicting the seizures by using multivariate statistical process control, which is widely known as an anomaly monitoring method. 90 seizures from the CHB-MIT EEG data of ten patients are analyzed.
The results of the proposed method demonstrated that 80 seizures out of 90 in preictal period were correctly predicted prior to the seizure onset, thereby giving a sensitivity of 88.89%. The false positive rate is observed to 0.39 per hour.
This study proposed a temporal based patient-specific epileptic seizure prediction method using MSPC in long-term scalp EEG signals. It also provides the possibility of realizing an EEG-based epileptic seizure prediction system which requires less computational power.
The proposed method does not require preictal data for modeling. The extracted features are computationally easy. The tested result shows good accuracy on the CHB-MIT data base.Le texte complet de cet article est disponible en PDF.
Multivariate statistical process control (MSPC) can be used for seizure prediction using EEG signal.
98% to less than 99% of variation of interictal data should be retained in the PCA model to predict seizures.
Less computational time because of temporal-based feature and MSPC.
Combination of ECG and EEG signal would be future work.
Keywords : Multivariate statistical process control, Squared prediction error (Q), Hotelling parameters, PCA, Epilepsy prediction