Prediction of epileptic seizure based on deep learning methods

Abstract

An epileptic seizure is a period when there is a rapid burst of intense electrical activity in the brain of a person. A person's behavior or movement might be change because of epileptic seizure attack. Patients may face different types of seizure. Different seizure attacks occur from different parts of brain. In many cases, seizure attack duration might be 30 seconds to less than two minutes. If a seizure attack cross the 5 minutes time duration then it will be a medical emergency case. Different circumstances can be the cause of seizure attack. The after effect of stroke can be a seizure attack. Severe head injury can be a cause of seizure attack. Any type of infection like as meningitis can be a cause for seizure attack. The specific facts that cause seizure are difficult to identify. Epileptic seizure can cause long lasting effect on human body like as hypertension, sleeping disorder etc.Prediction of epileptic seizure at the very early age can save an epileptic patient from these problems.In this research work,we proposed an epileptic seizure prediction method. In the model we proposed, we used Deep Learning model to predict epileptic seizure.We used Convolutional Neural Network(CNN), which is a section of deep neural network. Here Convolutional Neural Network is used for analysis the EEG signals. There are different phases of seizures. They are preictal, ictal, interictal. The seizures are sudden and unpredictable in nature. This is one of the most concerning aspects of epileptic seizure. The extraction of feature method and classi er using techniques are very much time consuming for classify ictal and interictal EEG signals. Deep learning can extract the features. In the dataset, there are epileptic EEG signals. The results from the procedures illustrate that proposed model can provide better performance over existing methods. Deep learning method used to make the model more time convenient.