A comprehensive EEG dataset and performance assessment for Autism Spectrum Disorder

Autism Spectrum Disorder (ASD) diagnosis can greatly benefit from more efficient and accurate tools to enable early intervention and reduce long-term healthcare costs associated with delayed diagnosis. Electroencephalography (EEG) has emerged as a promising non-invasive technique for detecting neural patterns linked to ASD. This research evaluates the effectiveness of three preprocessing techniques, Butterworth, Discrete Wavelet Transform (DWT), and Independent Component Analysis (ICA), in enhancing EEG signal quality for ASD classification. The performance of each method is assessed using Signal-to-Noise Ratio (SNR), Mean Absolute Error (MAE), Mean Squared Error (MSE), Spectral Entropy (SE), and Power Spectral Density (PSD) analysis to explore frequency band distribution. Additionally, Hjorth parameters—activity, mobility, and complexity—are computed to capture neural dynamics associated with ASD. Results showed that ICA achieved the highest SNR values (normal: 86.44, ASD: 78.69), indicating superior denoising capability, while DWT offered the lowest error metrics (MAE: 4785.08, MSE: 309,690 for ASD), demonstrating its robustness in preserving signal characteristics. Butterworth provided moderate results across metrics. Notably, Hjorth parameters revealed that neurotypical EEGs exhibited higher activity and complexity, highlighting distinct neural dynamics compared to ASD. These findings suggest that ICA is optimal for applications prioritizing signal clarity, while DWT offers a balanced approach for feature preservation in ASD EEG analysis. These findings are expected to support the development of more accurate, EEG-based diagnostic tools for ASD that can be integrated into clinical decision support systems and early screening programs.