Open AccessEpilepsi detection system based on EEG record using neural network backpropagation method
Author(s) -
Ade Eviyanti,
Hindarto Hindarto,
Sumarno Sumarno,
Herlian Aliyasa Alamj Duddin
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1381/1/012037
Subject(s) - backpropagation , epilepsy , electroencephalography , feature extraction , computer science , pattern recognition (psychology) , artificial intelligence , artificial neural network , signal (programming language) , data set , discrete cosine transform , set (abstract data type) , feature (linguistics) , epileptic seizure , speech recognition , psychology , neuroscience , image (mathematics) , linguistics , philosophy , programming language
Epilepsy is a manifestation of brain disorders with a variety of etiologies, but with the typical single symptom, periodic and reversible attacks, Epilepsy is characterized by an excess amount of electricity coming out of the brain cells, which can cause seizures and abnormal movements. EEG signals on epilepsy attacks have a characteristic pattern that allows health professionals to distinguish them from normal conditions (nonseizure). There are many methods used by researchers to recognize patterns of EEG epilepsy and non epilepsy signals in this study using Discrete Cosinus Transform (DCT) to perform the extraction of EEG signal features and Backpropagation Neural Networks for identification of EEG signal patterns. This research data using five classes of data sets of digital EEG signal taken from clinic Epileptologie University of Bonn ie data set A normal open eye signal, set B normal eye closed signal, C set enter the epilepsy zone, set D enter epilepsy, set E epilepsy seizures. The five class data are processed using DCT to obtain feature extraction, so results from DCT are used to perform identification using the Backpropagation method. The results of this study indicate that with feature extracted using DCT and identification process using Artificial Neural Network Backpropagation got EEG signal identification obtained for data set A, B, C, D and E is 76%, for set AB and CDE class data 73 %.