Epilepsy Disease Detection Using the Proposed CNN-FCM Approach

The electroencephalogram (EEG) signals are categorized using fuzzy C-means (FCM)-based deep learning classification method. The internal structure of this conventional CNN architecture is in sequential order, whereas the interior structure of the projected CNN architecture is in parallel order. The signals are initial data augmented of EEG samples in both cases are decomposed using empirical mode decomposition (EMD) transformation model in the existing methodology. The decomposed samples are given into the feature extraction process, where decomposed samples are trained. These computed IMF subbands and the intrinsic features are fed into the proposed CNN-FCM model to produce the trained sequences for the training phase of the network. In the testing model of the proposed system, the data augmentation process is applied to the test EEG signal, and then, EMD transformation model is applied to the data-augmented EEG samples to obtain the IMF subbands. The same architecture proposed in this work is also used for diagnosing the severity of focal EEG signals.