Abstract / Summary
Attention Deficit Hyperactivity Disorder (ADHD) is one of the most common neurodevelopmental disorders in children, making early detection and intervention critically important. In this study, we propose a novel EEG based classification framework for ADHD that integrates advanced nonlinear signal analysis with deep learning methodologies. EEG recordings were collected from 61 children diagnosed with ADHD and 60 age-matched healthy controls during a visual attention task, utilizing 19 electrodes placed according to the international 10-20 system. The preprocessing pipeline involved artifact rejection, independent component analysis (ICA), and band-pass filtering. A key innovation of our method is the integration of spherical phase space partitioning with entropy-optimized symbolic time series analysis (SPSP-STSA), allowing for reliable and noise-resilient feature extraction across multiple EEG channels. The extracted symbolic sequences were then used for classification via cosine similarity and a bidirectional Long Short-Term Memory (LSTM) network, which effectively models temporal patterns to improve diagnostic accuracy. The proposed method achieved a classification accuracy of up to 98% using window-based analysis of 20-second EEG segments. Our findings highlight the potential of SPSP-STSA and deep learning for advancing EEG-based ADHD diagnosis, offering improved robustness and interpretability compared to conventional approaches.