EEG Analysis of Neurodevelopmental Disorders by Integrating Wavelet Transform and Visual Analysis

Identifying neurodevelopmental disorders, ADHD, autism spectrum disorder, and other disorders (e.g., depression and mental health diseases), are important for planning appropriate treatments and early intervention. EEG is a commonly used method that measures the electrical activity of a brain to examine such disorders. This study introduced an approach to understanding the disorders by integrating wavelet transform and visual analysis on EEG signals. Wavelet-based features are extracted to find informative information associated with any changes in the EEG signals to differentiate them from healthy subjects. The effectiveness of the features is determineed by proposing two different feature selection methods (DWT-PCA and DWT-ANOVA) and evaluated by applying ML classification algorithms, such as KNN and Naive Bayes. Also, visual analysis is conducted to assess the features and to enhance the understanding of the features. We found that the classification with DWT-PCA features provided better performances. Although there was no clear distinction between normal (i.e., healthy) and abnormal (i.e., disorders), similarities and differences between them were identified through visualization. Overall, the integration of using both wavelet-based feature extraction and visual analysis was effective in identifying diagnostic neurodevelopmental disorders.