A Robust Computational Framework for Autism Spectrum Disorder Identification Using Optimized Image Processing and...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Autism Spectrum Disorder (ASD) classification is evolving with machine learning and image analysis.
• Current methods require robust preprocessing and feature extraction for accurate identification.

Purpose of the Study:

• To introduce a novel framework for enhanced ASD classification using advanced image processing and deep learning.
• To improve the accuracy and efficiency of early ASD detection.

Main Methods:

• Utilized Discrete Wavelet Transformation with Gaussian Filter Kernel (DWT-GFK) for robust image preprocessing.
• Employed the VGG16 deep learning model for feature extraction and pattern identification.
• Implemented an Elevated Learning-based Boosting Network with Hybrid Learning with Neural Classifier Logic (ELBN-HLNCL), optimized by the Walrus Optimization Algorithm (WaOA), for classification.

Main Results:

• Achieved exceptional performance with 99.95% accuracy, 99.9% recall, 99.8% precision, and 99.85% F1-score.
• Demonstrated superior classification results across multiple datasets including AID, ASD screening, and ABIDE.
• Outperformed existing ASD classification approaches in experimental testing.

Conclusions:

• The proposed DWT-GFK, VGG16, and ELBN-HLNCL framework offers a highly accurate and effective method for ASD classification.
• The findings support the development of improved diagnostic tools for early ASD detection.
• This research provides valuable applications for healthcare professionals in making accurate ASD assessments.