The Automatic Seizure Detection Method Based On Similarity

Epilepsy is one of the highest rate of the chronic brain disease. Epileptic seizures are clinical manifestation of abnormal and excessive neuronal discharges in the brain which will result in the functional disorder of the brain. The tradi-tional seizure detection is generally completed by the trained doctors according to the visual examination and clinical manifestation. However, the long-term electroencephalograms (EEGs) and the subjective process became two major is-sues on the seizure detection. Therefore, how to design the automatic seizure detection method has attracted more and more researchers in recent years, by which the seizure can be detected automatically with the application of machine learning methods and computer technology. This paper proposes a new auto-matic seizure detection method which is based on the dynamic-similarity-feature and extreme learning machine. The main contents are presented as follows:Chapter 1 systematically introduces the background and development of the automatic seizure detection, including the concepts of epilepsy and EEG, as well as the basic process of automatic seizure detection.Chapter 2 mainly discusses two key topics in the automatic seizure detec-tion, that is, feature extraction and classification. The similarity-based feature extraction methods and the complex-based feature extraction methods are first introduced respectively. Then three classifiers, artificial neural network, extreme learning machine and support vector machine, are presented.Chapter 3 first proposes a new dynamic similarity-based feature extraction method. Then combining extreme learning machine, an automatic seizure de-tection method are constructed. Finally, the feasibility and efficiency of the proposed method are verified on Bonn data base and CHB-MIT data base.