A Smart Integrated Brain-Computer Interaction for Epileptic Seizure Detection

Epilepsy is brain resulted activities which are affected by suddenly seizures which have unpredictable changes affects brain electrical functionalities. Epilepsy has a significant impact on society on the healthcare treatment, cost, responds, and patients behavior. The study has main objectives to propose accurate integrated framework for epileptic seizure detection from the pre-ictal phase of the EEG signal. Locate the connected channel lobe in region where epileptic is expected to occur. Provide automated and real-time monitoring and send warning messages to patient and epileptologist to take accurate actions before ictal occur. Enable future contribution for different Seizure features and impact. Also reduce cost, time and effort. Based on the hypothesis of entropy of EEG signals during seizure has low value if (n) of channels are detected to have seizure, then they are considered as connected neighbors in brain domain mapping, which is clear alert that patient will have a seizure ictal. This end to end framework has modules of data acquisition, pre-processing, feature extraction, pattern-matching, supports vector machines (SVM) classifier for extracted feature, in addition to monitoring and notification. The extracted features includes lower threshold, homogeneity, weighted permutation entropy, power and energy. Also identify the physiological field located inside the brain which the seizure will expected to occur. The final output results have 92% for True positive rate in addition to 95% of F1 and 98.9% of accuracy. This system has proved consistency during all its phases of seizure detection with valuable and effective support to the society.