Classification of Non-focal and Focal EEG signals using Local Binary Pattern

Abstract

Electroencephalogram is a clinical diagnoses tool that monitors the electrical impulses of the cerebrum. It is used to sense the glitch in the brain due to the recurrent existence of the seizures known as epilepsy. The detection of epileptic seizures by human examination is time consuming and it results in misconception. Therefore in this paper an effective feature extraction method of local binary pattern (LBP) is introduced for the automatic identification of epilepsy to reduce the complexity of the human examination. The extracted features are classified by employing artificial neural network (ANN) classifier to discriminate non-focal and focal EEG signals. The epilepsy EEG dataset furnished by Bern- Barcelona contains 3750 pairs of EEG signals from non-focal and focal class used in this study. 10-fold cross validation is performed to evaluate the discrimination performance. The proposed method LBP with ANN classifier achieved a 93.21% of accuracy, 93.63% of specificity and sensitivity of 92.80%.