Development of EEG epilepsy recognition system using artificial neural network

Yahya, Adilah (2021) Development of EEG epilepsy recognition system using artificial neural network. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

	Text (24 Pages) Development of EEG epilepsy recognition system using artificial neural network.pdf - Submitted Version Download (613kB)
	Text (Full Text) Development of EEG epilepsy recognition system using artificial neural network.pdf - Submitted Version Restricted to Repository staff only Download (9MB)

Abstract

Epilepsy classified as one of the brain-related disorders affecting the entire brain nervous system and characterized by the high-frequency and high-voltage brain waves, which called as a seizure. This disease recognized as one of the uncontrollable movements made by the epilepsy patients during an outbreak which can cause consciousness and convulsion. Consequently, this thesis studies are to develop EEG Epilepsy Recognition System using Artificial Neural Network (ANN). Cascade-Forward Neural Network technique is used as their primary tools which their system invented to execute a process like a human brain. This brain-inspired system which meant to recreate the way human brains thinks. This thesis uses Cascade-forward Neural Network, which is quite similar to the Feed-forward Neural Networks approach and proposes the epilepsy recognition procedure implemented by using MATLAB software. Additionally, this study also used the Electroencephalogram (EEG) signal which then used to diagnose and accessing human brain activity and disorder by using the dataset obtained from University of Bonn (UBonn), which has been widely used by other researchers regarding epilepsy studies. The MindLink EEG Sensor is used to obtained external data of the EEG signal, which then needed to test in the neural network. As for results, this system successfully carried out with 79.4% for overall accuracy with training, validation and testing sequentially acquire 79.4%, 88.0% and 70.3%.

Actions (login required)

View Item