Epileptic Seizure Classification and Prediction Model Using Fuzzy Logic-Based Augmented Learning

Introduction

Epilepsy is a chronic disorder caused due to the presence of seizures. It is one of the brain disease caused by the imbalance of flow of electron in the brain. There is a sudden bust in the electrons which leads to a high electricity in the brain. It can be in common said to be as a jerk which will lead to act a human abnormal from normal conditions. Meditation, long term or short term drugs therapy is the primary treatment for the epileptic seizures which is given under the condition of temporal lobe but when the brain doesn’t react on this then there is a chance of secondary treatment where the frontal lobe occurs and there is a call for the urgent brain surgery. The risk of successfully removing the tissues from the part of brain which is affected with seizures is comparable high. To avoid the risk and rate of sudden death it is important to predict the seizures before it could be onset.

EEG is the methodology used to record and keep a continuous update of brain signals. There are different techniques like Magneto encephalography (MEG), Functional magnetic resonance imaging (fMRI), Electro Encephalography (EEG) and computerized tomography (CT) and rest more. But EEG is preferred by many of the researcher in their research due to its adoptive nature. As the signals obtained from the EEG are disturbed and non-continuous, there is a need for designing model with preprocessing of signal. There are usually two categories of EEG signals, iEEG and scalp EEG signals. It is basically placement of the electrodes if the electrodes are placed on the scalp of human then it is known as scalp EEG signal and signals which are obtained from during the brain surgery there are known as iEEG signals. EEG signals are very sensitive they need to be strengthened by doing signal processing were the filters are used to remove the undesired signals. Epileptic seizures can be attacked at any instant of time. There are effected by both the gender’s and there is no age limit for this. It can be genetic disorder for some special cases in the children or it can be caused due to high fever and brain damage also. It’s a difficult task to predict when epileptic seizures will occur. So there by need for continuous monitoring of brain signal is mandatory for epileptic seizures patients. EEG is the used for measuring and monitoring brain signal.

Figure 1.

EEG signal with stages of seizures

The Figure. 1 shows the clear idea of the different stages of seizures in the EEG signal. Onset is the time instant where the seizures are initiated from normal waves. The normal to abnormal signal disturbance can be clearly identified in Ictal state. Inter-ictal is the condition from one seizure to another seizure. Ictal is the actual state of seizures where the attack occurs, pre-ictal is the state before the seizures attack and post ictal is the state after the seizures attack. Seizures are expected to be controlled by the closed loop strategies rather than open loop strategies. The algorithm in this paper is designed to predict the seizures onset before it could occur.

The current drawback in aligning the research challenges is to provide a sustainable ecosystem for detecting and classifying the seizure occurrence using fuzzy and augmented intelligence approach. The datasets (EEG) is streamlined with a trivial pattern for classifying and detecting a predictive occurrence of seizure and hence the proposed solution provides a feasible approach for customizing datasets and patterns. The manuscript is organized with a brief survey on current literature reviews in section II followed by the proposed methodology in section III. The training data repository classification and mathematical proof is discussed in section IV with the results and discussion from the experimental setup is demonstrated in section V, followed by a conclusion.