Diagnosis of epilepsy from interictal EEGs based on chaotic and wavelet transformation

• DOI: 10.1007/s10470-016-0810-5
• Publication Date: 2016-08-02T02:29:47Z
• Authors: Jisu Elsa Jacob; Vijith Vijayakuma...; Thomas Iype; Gopakumar Kuttapp...; Doris George Yoha...
• Keywords: Correlation dimension

In this study, we have reinvestigated the chaotic features and sub-band energies of EEG and its ability for aiding neurologists in detecting epileptic seizures. The study was done on the EEG of ictal and interictal phases of epileptic patients and of normal subjects. The EEG was decomposed using discrete wavelet transform to obtain various sub-bands and the chaotic features like correlation dimension and largest Lyapunov exponent were extracted. The analysis results clearly show that the correlation dimension and largest Lyapunov exponent have their lowest value during seizure activity, higher for interictal and even higher values for normal EEG. These values strongly suggest that interictal phase EEG of an epileptic patient is less complex and more predictable compared to normal EEG. Chaotic features extracted are potential parameters for automated diagnosis of epilepsy. Support vector machine (SVM) classifier was implemented based on both sub-band energies and chaotic features extracted from EEG. Classification performance parameters of SVM classifier based on sub-band decomposed energies and chaotic features were calculated.