Real-Time Epileptic Seizure Detection from EEG Signals via Random Subspace Ensemble Learning

Real-time detection of seizure activities in epileptic patients is crucial and can help improve patients' quality of life. Accurate evaluation, pre-surgery assessments, seizure prevention, and emergency alerts for medical aid all depend on the rapid detection of the onset of seizures. A new method of feature selection and classification for rapid and precise epileptic seizure detection is discussed. In this solution, informative components of Electroencephalogram (EEG) data are extracted and an automatic method is presented using Infinite Independent Component Analysis (I-ICA) to select efficiently independent features. The feature space is divided into subspaces via random selection, and multi-channel Support Vector Machines (SVMs) are used to classify the subspaces, then, the result of each classifier is combined by majority voting to find the final output. To evaluate the solution, a benchmark clinical intracranial EEG (iEEG) of eight patients with temporal and extratemporal lobe epilepsy has been considered in a multi-tier cloud-computing architecture. Via the leave-one-out cross-validation, accuracy, sensitivity, specificity, and false positive and false negative ratios of the proposed method are 0.95, 0.96, 0.94, 0.06, and 0.04, respectively, which confirm the effectiveness of the proposed solution.