Article Preview
Top1. Introduction
Epilepsy is a central nervous system (neurological) disorder that causes abnormal brain activities (Gupta et al., 2020). It is the 4th neurological disorder affecting roughly 50 million people around the world according to the World Health Organization (WHO) report (https://www.who.int/mental_health/neurology/epilepsy/en/). The abnormal brain activities called epileptic seizures suddenly attack the patient and threaten his\her life. These seizures may occur in a specific area in the brain or include the whole brain. The seizure symptoms are related to the position and the extension of the abnormal brain activities. Some patients may harm themselves because they lose control on their muscular activities (Gajic et al., 2014). Hence, any serious steps to diagnose and treat this disorder are urgently required.
Electroencephalogram (EEG) is an essential diagnostic tool in neurological clinics due to its ability to record the electrical brain activity (Singh, 2019). These records are a worthy source of information about epileptic seizures (Janghel et al., 2019). Seizures show rapid strong fluctuations in EEG signals.However, due to the unpredictability nature of epileptic seizures, 24-hour monitoring is usually needed which is a strenuous and time-consuming process(Mardini et al., 2020). Thus, a robust automatic EEG based epilepsy detection system is highly demanded. For clinical purposes, this system should have the ability to discriminate between three classes: normal, interictal (during seizure free time interval) and ictal (during a seizure).
The initial steps toward such system began in the second half of the twentieth century. Later, many studies (Banerjee et al., 2019;Gajic et al., 2014; Guo et al., 2010a; Guo et al., 2010b; Husain & Rao, 2012;Ibrahim & Majzoub, 2017; Juarez-Guerra et al., 2015; Mammone et al., 2015; Nanthini & Santhi, 2017; Redelico et al., 2014;Srinivasan et al., 2007;Subasi, 2006; Tawfik et al., 2016; Wang et al., 2011; Wang et al., 2017; Wei et al., 2019)have been published with promising accuracy. However, most of these studies ignore the interictal class or sometimes it is considered as a normal class (Guo et al., 2010a; Guo et al., 2010b; Tawfik et al., 2016). This can be interpreted by the difficulty of distinguishing an interictal class from the others.