Wavelet Based Multi-Class Seizure Type Classification System

Epilepsy is one of the most common brain diseases that affect more than 1\% world's population. It characterized by recurrent seizures, which come in different types and are treated differently. Electroencephalography (EEG) commonly used medical services to diagnose seizures their types. The accurate identification helps provide optimal treatment information patient. However, manual diagnostic procedures epileptic laborious highly-specialized. Moreover, EEG evaluation a process known have low inter-rater agreement among experts. This paper presents novel automatic technique involves extraction specific features from signals using Dual-tree Complex Wavelet Transform (DTCWT) classifying them. We evaluated proposed on TUH Seizure Corpus (TUSZ) ver.1.5.2 dataset compared performance with existing state-of-the-art techniques overall F1-score due class imbalance seizure Our achieved best results weighted 99.1\% 74.7\% for seizure-wise patient-wise classification respectively, thereby setting new benchmark this dataset.