Automated epileptic seizures detection using multi-features and multilayer perceptron neural network.

摘要: Detection of epileptic seizure activities from long-term multi-channel electroencephalogram (EEG) signals plays a significant role in the timely treatment patients with epilepsy. Visual identification EEG is cumbersome and tedious for neurologists, which might also lead to human error. Therefore, an automated tool accurate detection seizures essential clinical diagnosis. This study proposes algorithm using multi-features multilayer perceptron neural network (MLPNN) classifier. After appropriate approval ethical committee, recordings data were collected Institute Neurosciences, Ramaiah Memorial College Hospital, Bengaluru. Initially, preprocessing was performed remove power-line noise motion artifacts. Four features, namely power spectral density (Yule–Walker), entropy (Shannon Renyi), Teager energy, extracted. The Wilcoxon rank-sum test descriptive analysis ensure suitability proposed features pattern classification. Single fed MLPNN classifier evaluate performance study. simulation results showed sensitivity, specificity, false rate 97.1%, 97.8%, 1 h−1, respectively, multi-features. Further, indicate suitable real-time recognition recording. graphical user interface developed MATLAB provide biomarker normal signals.