Detecting epilepsy in EEG signals using synchro-extracting-transform (SET) supported classification technique

Epilepsy is one of the medical conditions in human caused by the disorder in central-nervous-system (CNS). Early detection and treatment are essential patient healthcare. Brain condition monitoring with electroencephalogram (EEG) is a commonly adopted medical practice as it provides vital information regarding the brain activity. This research work aims to present a detailed examination on EEG signals with improved detection accuracy using synchro-extracting-transform (SET); which converts the complex 1D EEG into 2D images using time-frequency transformation. The proposed EEG classification pipeline consists of the following phases: (i) transforming the EEG into RGB scaled image, (ii) implementing the discrete-wavelet-transform and local-binary-pattern to enhance the image textures, (iii) mining the essential texture and entropy features, (iv) dominant feature selection with firefly algorithm, (v) serial feature concatenation, and (vi) binary classifier implementation and fivefold cross validation. In this work, the classification of EEG is performed with; (i) handcrafted features, (ii) deep-features and (iii) concatenated deep and handcrafted features and the results were presented and discussed. The proposed approach helped to achieve a better classification accuracy with handcrafted features, deep features and concatenated features for EEG datasets considered in this research.