2020 Volume 48 Issue 3 Pages 107-112
Computer-aid system using machine learning for epileptic focus detection is having a promising future. In this study, we report some initial results on machine learning based automatic identification of epileptic focus from intracranial EEG (iEEG) data. We analyzed 90 segments from 30 minutes of iEEG from four surgical patients with focal cortical dysplasia. Frequency band were divided into δ, θ, α, β, γ, ripple, and fast ripple. Subsequently, the prominent entropies were used and calculated mutual information value from the following; Appropriate, Sample, Permutation, Shannon, Renyi, Tsallis, and phase 1 & 2. Finally, support vector machine was used for classifying the epileptic from non-epileptic foci. The accuracy of our method was evaluated by 10-fold cross validation. Appropriate and sample entropies were proper to distinguish the epileptic electrodes. Automated detection using our method resulted higher AUC from 0.63 to 0.83. Machine learning using the multiband entropy-based feature-extraction method is useful for epileptic focus detection.
