Development of epileptic seizures detection based on EEG signals analysis using MATLAB

Abdul Latef, Mohamad Aiman (2021) Development of epileptic seizures detection based on EEG signals analysis using MATLAB. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Development of epileptic seizures detection based on EEG signals analysis using MATLAB.pdf - Submitted Version Download (2MB)

Abstract

Brain epilepsy is classified as a brain-related disorder that affects the entire brain's nervous system and is caused by brain waves undergoing high-frequency voltage changes or seizures. The disease is one of the uncontrolled movements performed by unconscious epilepsy patients. Therefore, the purpose of this study is to develop the detection of epilepsy seizures based on EEG signal analysis using Matlab. For results, this study initially used the feed-forward neural network technique. The feed-forward neural network technique is used in machine learning to execute the exact processes of the human brain. The system was inspired by the way the human brain thinks. This study uses input data such as standard deviation and mean, and then this epilepsy detection procedure is implemented using MATLAB software. Testing was performed sequentially from 10 to 20 neurons to obtain the best values. In addition, this study also uses an electroencephalogram (EEG) to diagnose and access activity in the human brain using a data set obtained from the University of Bonn (UBonn), which other researchers in this epilepsy study have widely used. A MindLink EEG sensor was used, and the sensor sent data via PLX- DAQ software to detect healthy or epilepsy. The standard deviation provides the best overall accuracy compared to mean, which is 80.8% at seventeen neuron while mean highest overall accuracy only 72.6%. Then, a Matlab Graphical User Interface (GUI) is used to execute editor Matlab file by pressed the pressed the button. GUI acts as shortcut keys to perform a task that takes couple of actions.

Actions (login required)

View Item