Review: A Survey of performance and techniques for automatic epilepsy detection

Abstract

Epilepsy is a chronic neurological disorder of the brain that affects around 50 million people worldwide. The early detection of epileptic seizures using electroencephalogram (EEG) signals is a useful tool for several applications in epilepsy diagnosis. Many techniques have been developed for unscrambling the underlying features of seizures present in EEGs. This article reviews the seizure detection algorithms developed in the last decade. In general terms, techniques based on the wavelet transform, entropy, tensors, empirical mode decomposition, chaos theory, and dynamic analysis are surveyed in the field of epilepsy detection. A performance comparison of the reviewed algorithms is also conducted. The needs for a reliable practical implementation are highlighted and some future prospectives in the area are given. Epilepsy detection research is oriented to develop non-invasive and precise methods to allow precise and quick diagnoses. Finally, the lack of standardization of the methods in the epileptic seizure detection field is an emerging problem that has to be solved to allow homogenous comparisons of detector performance.