부산대학교 도서관

Epilepsy has been derived from the Greek word meaning ‘seizure’. It means sudden at-tack in medical terms. It is a neurological disorder having sole features and the tendency of recurrent seizures. The main foundation of seizures is when the brain encounters electrical activity disturbance; the main manifestation of epilepsy is encountering more than one seizure in the span of 24 hours. As a result, a thorough change in the patient's behavior or senses may occur and the patient may remain in an unconscious state for a while because of the unexpected nervous breakdown generated in the brain as a result of the seizure. In recent years researchers applied Deep Learning technique to discover sensible and meaningful patterns from different domain datasets to predict the seizure attack in early stage. Applications of ML can also be seen on brain datasets for seizure detection, epilepsy lateralization, differentiating seizure states, and localization of the Epileptic Zone (EZ). However Deep Learning techniques provide the potential for analyzing the datasets precisely and in fast manner that is possibly not discovered in past. In this article, we conducted a comparative study of different contemporary techniques available in the literature for seizure detection. These techniques use a combination of various medical tests available for the detection of epilepsy which is EEG, MRI, fMRI, PET, and rs-fMRI. Identification of better combinations of these tests can predict better results using Ma-chine Learning techniques that ultimately helps the epileptic patient for better medication. On the observations made after the literature review in contemporary research articles, it has been identified that there are still many aspects related to the detection of epileptic disease that are unexplored using current computational learning techniques.