| Statistical EEG Seizure Prediction for Brai Computer Interfaces. |
| Paper ID : 1155-ICEEM2023 (R1) |
| Authors |
|
Prof. Fathi Sayed *1, Mostafa El-sayed2 1Faculty of Electronic Eng 2Communication , menouf faculty , menoufia university |
| Abstract |
| Nearly 50 million people worldwide suffer from epilepsy, one of the most prevalent neurological disorders. Epilepsy is clearly reflected in EEG signals. These signals are of multi-channel nature, and are acquired with electrodes that are put on the human head. Epilepsy patients suffer from unexpected seizures. Generally, unexpected seizures involving involuntary contractions of the muscles and loss of consciousness are frequently brought on by epilepsy. All epilepsy patients suffer greatly from the unpredictable nature of the disease's development. In this paper, a proposed wavelet-based method for EEG channel selection and subsequent epilepsy seizure prediction using straightforward statistics is presented. The signal is divided into small segments. Five signal attributes are extracted for each signal. These attributes are treated as random variables for all signal activities. The discrimination between normal and pre-ictal activities is the process of prediction. This is performed based on a statistical approach. PDFs for each activity are determined, and intersection points are set for prediction. For each new incoming segments, a thresholding process is performed to take the decision. A majority voting strategy is performed after that to allow a robust decision about the segment signal activity for efficient seizure prediction. |
| Keywords |
| EEG, Seizure prediction, channel selection |
| Status: Accepted |