EEG Based Autism Detection Using CNN Through Correlation Based Transformation of Channels' Data
Zahrul Jannat Peya, Zahrul Jannat Peya, M. A. H. Akhand, M. A. H. Akhand, Jannatul Ferdous Srabonee, Jannatul Ferdous Srabonee, Nazmul Siddique, N. Siddique
Abstract
Early diagnosis of autism or autism spectrum disorder (ASD) can help improving behavioral, language development and communication skill. As ASD is a neurodevelopmental disorder, brain signals are used to early diagnosis. Among different brain signals, electroencephalography (EEG) is the effective one. Electrical brain activity is measured through EEG signal from the scalp of human over a period of time and is used to analyze complex neuropsychiatric problems of brain. This study investigates an EEG based ASD detection using CNN, the well-known deep learning method for image analysis and classification. At first, the individual EEG channel data are transformed into 2D form using Pearson's Correlation Coefficient and then classification is carried out using the well-known CNN model residual neural network. Experiments performed on clinical EEG data show that the proposed approach achieved a classification accuracy of 100%.