Epileptic Seizure Detection Using Machine Learning Techniques

Introduction

There are various definitions for epilepsy and epileptic seizure in the literature. The International League Against Epilepsy (ILAE) and the International Bureau for Epilepsy (IBE) took a joint decision on the descriptions of epilepsy and epileptic seizure. In the sequel, they described these terms as “a disorder of the brain characterized by an enduring predisposition to generate epileptic seizures and by the neurobiologic, cognitive, psychological, and social consequences of this condition” for epilepsy and “a transient occurrence of signs and/or symptoms due to abnormal excessive or synchronous neuronal activity in the brain” for epileptic seizure. Besides, at least one epileptic seizure must occur to describe epilepsy (Fisher et al., 2005).

All over the world, epilepsy is one of the most widespread neurological illnesses and many people of different races, geographical areas, and ages suffer from this disease. Globally, almost 50 million people have epilepsy disease and approximately 80% of them live in low- and middle-income countries (WHO, 2019; WHO, 2020). As seen from Figure 1 indicating worldwide idiopathic epilepsy age-standardized prevalence per one hundred thousand population among males and females in 2016, idiopathic epilepsy is the most common in southern, western, and eastern sub-Saharan Africa, Andean and central Latin America, and central and southeast Asia (Feigin et al., 2019; WHO, 2019).

Figure 1.

Worldwide idiopathic epilepsy prevalence per one hundred thousand population in 2016

(Feigin et al., 2019; WHO, 2019)

Timely diagnosis of the occurrence of epileptic seizures is one of the basic difficulties. The epileptic seizure detection process is fundamentally carried out by medical doctors with reading the electroencephalogram (EEG) recordings because of the unpredictable and complex nature of the illness, taking much time and depending on the experts’ practice. Consequently, automated seizure detection using EEG signals is required and crucial for the comfort of medical specialists and patients (Yavuz, Kasapbaşı, Eyüpoğlu, & Yazıcı, 2018). Machine learning (ML) techniques are utilized while detecting epileptic seizure automatically. This chapter describes how ML methods are utilized to detect epileptic seizures and investigates the existing studies in the literature. Furthermore, the basic aim and contribution of this chapter are to inform the researchers who will work in this area.

The rest of the chapter is organized as follows. The second section expresses how epileptic seizures are detected and how epilepsy disease is diagnosed. In the third section, existing ML techniques utilized for epileptic seizure detection are investigated. Finally, the fourth section concludes the chapter.

Detecting Epileptic Seizures And Diagnosing Epilepsy

Epilepsy disease is described with repetitive seizures that are short episodes resulted from extreme electrical discharges in some brain cells. Various sections of the brain may become the place of these discharges. Moreover, frequency of seizures ranges from less than one per year to several per day. As a result of these seizures, different types of nonpermanent symptoms may occur, such as loss of the senses of hearing, sight and taste, and defects of movement, consciousness or other cognitive functions. Patients with epilepsy suffer from physical problems including injuries and fractures related to epileptic seizures, and psychological consequences regarding depression and anxiety (WHO, 2020).

In order to early detect the disease and manage a variety of etiologies, it is vital to analyze the causes of epilepsy. The causes of epilepsy disease are split into six types which are genetic, structural, immune, infectious, metabolic and unknown, as demonstrated in Table 1 (Scheffer et al., 2017; WHO, 2019).