A study on deep learning approach based detection of seizures from eeg data

Abstract

Seizures can be described as the sudden change in a brain s electrical activity and it affects nearly fifty millions of the people worldwide. It is a rapid, temporary and unpredictable change related to the human brain s electrical activity and affects the people of all age groups. Generally, seizures are categorized into two main types: (i) focal seizure which affects only a particular part in the brain and (ii) generalized seizure which affects the entire part of the brain. newlineA person who is affected with focal seizures could survive without losing consciousness and may feel muscle contractions after relaxations. The generalized seizure is a dangerous seizure which makes the person affected to lose his or her consciousness and balance of mind and body. In addition, the smacking lips, rolling eyes and biting of tongue would occur in the affected person. newlineThe Electroencephalogram shortened as EEG is the generally available tool which can record the brain s signal called electrical activity. The EEG signals are recorded by placing the electrodes on the patient s scalp. The electrodes are fixed using two different methods. The first method is intracranial where the electrodes are implanted in the inside part of the scalp and the second method is extra-cranial where in the electrodes are fixed in the outside part of the scalp. newlineTraditionally, the EEG data have been analyzed manually where the neurologists diagnose the epilepsy by looking into the EEG recordings thoroughly for some amount of time. This is a time consuming and laborious task. There is a high chance for misdiagnosis of seizures in this case. To help the epileptic patients and to assist the neurologists in detecting seizures, newline