Hybrid groupers-moray orangutan optimization algorithm-based dense-resolution network for autism spectrum disorder
Urtti Bhagyalatha1, Bidush Kumar Sahoo1, Satish Muppidi2
1School of Engineering and Technology, Department of Computer Science and Engineering, GIET University, Gunupur, Odisha, India.
Psychiatry Research. Neuroimaging
|April 10, 2026
Summary
This study introduces a new AI model, GMOA-DRNet, for diagnosing Autism Spectrum Disorder (ASD) using brain scans and patient data. The model shows high accuracy, improving early detection and intervention for ASD.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Biomedical Engineering
Background:
- Autism Spectrum Disorder (ASD) presents significant cognitive and social challenges, necessitating precise and early diagnosis.
- Current diagnostic methods for ASD often struggle with accuracy and generalization, particularly with complex neuroimaging and behavioral data.
- Subtle patterns in brain imaging and phenotypic data are crucial for effective ASD diagnosis, but are often missed by conventional techniques.
Purpose of the Study:
- To develop an advanced diagnostic tool for Autism Spectrum Disorder (ASD) by integrating multimodal data.
- To enhance the accuracy and reliability of ASD detection using a novel optimization algorithm and deep learning network.
- To overcome the limitations of conventional diagnostic approaches in identifying subtle patterns in neuroimaging and behavioral datasets.
Main Methods:
- A multimodal approach combining Magnetic Resonance Imaging (MRI) brain images and autism-related phenotypic data.
- Utilized the Groupers and Moray Orangutan Optimization Algorithm (GMOA) for pre-processing, region identification, and network training.
- Employed feature extraction techniques (Tamura, GDP), data normalization, feature selection (ANOVA), and data augmentation (ADASYN) within a Dense-Resolution Network (DRNet).
Main Results:
- The proposed GMOA-DRNet achieved high diagnostic performance, with an accuracy of 92.77%.
- Demonstrated strong sensitivity (92.52%) and specificity (92.77%), indicating reliable identification of ASD cases.
- Achieved a low False Omission Rate (FOR) of 0.080, suggesting a minimal rate of missed diagnoses.
Conclusions:
- The GMOA-DRNet model offers a significant advancement in the accurate and reliable detection of Autism Spectrum Disorder (ASD).
- The multimodal data integration and novel optimization algorithm contribute to improved diagnostic performance over conventional methods.
- This AI-driven approach holds promise for enhancing early ASD diagnosis, facilitating timely intervention and improving patient outcomes.
