Continuous Wavelet Transform Spectral Image Based Epileptic Episode Classification

Numerous individuals globally, regardless of age, have the neurological condition epilepsy. Recurrent seizures compromised motor and sensory abilities, and a hindered normal lifestyle are all signs of epilepsy. By seeing or manually examining electroencephalogram (EEG) data, neurologists can make the diagnosis of epileptic seizures. In this paper, a continuous wavelet transform spectral image-based epileptic episode classification technique is proposed. The EEG signal was first converted into 227x227 RGB images using CWT. Then these images were used to create an image dataset, and classification was performed using the proposed deep ensemble epileptic seizure classifier (DE_ESC). DE_ESC ensembled Dense_net_169, Dense_net_201, and Inception_V3. This proposed method obtained 100% accuracy (Ac), precision (Pr), recall, and F1 Score, respectively.