Classification of the interictal state with hypsarrhythmia from Zika Virus Congenital Syndrome and of the ictal state from epilepsy in childhood without hypsarrhythmia in EEGs using entropy measures

This paper intends to classify the interictal state with hypsarrhythmia in patients with Zika Virus Congenital Syndrome (ZVCS) and of the ictal state in patients with epilepsy in childhood without the presence of hypsarrhythmia. Hypsarrhythmia is a specific interictal chaotic morphology, and the correct distinction between these two EEG states is crucial to improving the cognitive development of these epileptic patients. The proposed approach was assessed using the proprietary database of Casa Ninar, which contains data regarding children from northeastern Brazil born with microcephaly caused by the Zika virus. We also used data from the CHB-MIT database. Fundamental rhythms of the EEG signal δ, θ, α, and β were analyzed, and then decomposed by Discrete Wavelet Transform, in which 45 mother wavelet functions were tested to determine the most appropriate function to represent the EEG signals in the hypsarrhythmia interictal and ictal states. We extracted Shannon, Log Energy, Norm, and Sure entropy measures of the subbands as relevant features, and the combinations among them were applied in the state-of-the-art machine learning methods. The combination of Sure entropy with Shannon entropy, or with Log Energy and Norm, extracted from the δ rhythm, allowed for the best linear separability between the classes in most of the classifiers, obtaining 100% accuracy, sensitivity, and specificity.

Keywords: Discrete wavelet transform; Epilepsy in early childhood; Epileptic seizure; Hypsarrhythmia; Ictal; Interictal; Machine learning; Rhythms of the EEG Signal.