Detection of Epileptic Seizures using Convolutional Neural Network
Surbhi Gupta, Mustafa Sameer, Neeraj Mohan
Abstract
One of the most prevalent neurological ailments, Epilepsy, affects around 1-2% of the entire population of earth. It is the second only of stroke when it comes to neurological sickness. The excessive and hypersynchronous activity of neurons in the brain is occurred due to the unanticipated breakdown and synchronization of a set of neurons in the brain leads to an epileptic seizure. Most neurologists widely use Electroencephalogram (EEG) signals to identify epilepsy by recording the brain's electrical activity directly. Nonetheless, for recording long EEG, the visual interpretation turns out so an intensive, expensive, and tedious error-prone exercise. Therefore, there is an ever-growing requirement for developing an effectual method for detection of automatic seizure. The author proposed a lightweight CNN architecture for seizure classification. High accuracy is achieved in only 20 epochs with few trainable parameters for binary classification.