Epileptic Seizure Detection Using CNN

Epilepsy is one of the most devastating diseases in the history of mankind. A neurological disorder in which irregular transmission of brain waves result in seizures and physical and emotional imbalance. This paper presents the study of effective use of Convolutional Neural Network (CNN) a deep learning algorithm for epileptic seizure detection. This innovative technology will help the real world to make diagnosis of the disease faster with greater accuracy. A binary class epilepsy dataset is considered with 179 feature extracted as the attributes. Binary and multiclass classification is performed by considering the class with epileptic activity against all non-epileptic class. For classification CNN model is proposed which achieved an accuracy of 98.32%. Then by applying a different multiclass data set having 4 classes, the degree of generalization of our model is also checked as the accuracy of end results was high. For 10-fold cross validation and 70–30 data splitting method our models for performance is evaluated using various performance metrics.