Automatic seizure detection using neutrosophic classifier
- PDF / 2,638,987 Bytes
- 10 Pages / 595.276 x 790.866 pts Page_size
- 12 Downloads / 225 Views
SCIENTIFIC PAPER
Automatic seizure detection using neutrosophic classifier Abdul Quaiyum Ansari1 · Priyanka Sharma1 · Manjari Tripathi2 Received: 15 October 2019 / Accepted: 12 July 2020 © Australasian College of Physical Scientists and Engineers in Medicine 2020
Abstract Seizures are the most common brain dysfunction. EEG is required for their detection and treatment initially. Studies proved that if seizures are detected at their early stage, proper and effective treatment can be given to patients. Automatic detection of seizures using the EEG signal was a very powerful area of research during the last decade. Various techniques have been proposed in the literature for feature extraction and classification of recorded EEG signals for seizure detection. However, to achieve reliable performance, some challenges in this area need to be addressed. In this work, an algorithm for seizure detection has been proposed, which is a combination of frequency-domain features and neutrosophic logic-based k-means nearest neighbor (NL-k-NN) classifier. An EEG database, collected at All India Institutes of Medical Sciences (AIIMS), New Delhi, has been used to test the performance of the proposed algorithm. The consistency in the performance of the proposed algorithm has been checked by applying it to the well-known Bonn University and CHB-MIT scalp EEG datasets. The classification accuracies of 98.16%, 100%, and 89.06% were achieved when the proposed algorithm was tested with AIIMS, Bonn University, and CHB-MIT datasets, respectively. The main contribution of this study is that a novel neutrosophic classifier is proposed in the field of seizure detection, for improvement in reliability and precision. The accuracy of the NL-k-NN classifier has further been established by comparing it with the reported results of linear discriminant analysis (LDA), support vector machine (SVM), and traditional k-NN classifiers. Keywords Neutrosophic logic · k-NN classifier · Seizure detection · Alpha-delta ratio (ADR)
Introduction Epilepsy is a neurological dysfunction with which people are suffering worldwide. Recurrent seizures occur in patients suffering from this disease. The human brain is made up of several neurons and when these neurons are hyperactive in a synchronized manner, seizures occur. If seizures are not controlled at its early stage, it may cause a dangerous situation for epileptic patients. Visual analysis of electroencephalogram (EEG) by the neurologist is a common practice for the * Priyanka Sharma [email protected] Abdul Quaiyum Ansari [email protected] Manjari Tripathi [email protected] 1
Department of Electrical Engineering, Jamia Millia Islamia, New Delhi, India
All India Institute of Medical Sciences (AIIMS), New Delhi, India
2
diagnosis of seizures, because of variability in EEG patterns at different stages viz., before a seizure (preictal), during the seizure (ictal) and after the occurrence of the seizure (postictal). But this traditional analysis method becomes very exhaustive as EEG recordings a
Data Loading...
