Sparse Multiview Task-Centralized Ensemble Learning for ASD Diagnosis Based on Age- and Sex-Related Functional...

Area of Science:

• Neuroscience
• Computer Science
• Medical Imaging

Background:

• Autism spectrum disorder (ASD) is a neurodevelopmental disorder impacting brain functional connectivity (FC).
• ASD-related FC alterations exhibit age- and sex-specific patterns.
• Current diagnostic methods often overlook these demographic variations.

Purpose of the Study:

• To develop an advanced computer-assisted diagnostic method for ASD.
• To incorporate age and sex information into the classification process.
• To improve the accuracy of image-based ASD diagnosis.

Main Methods:

• Proposed a novel sparse multiview task-centralized (Sparse-MVTC) ensemble classification method.
• Formulated ASD diagnosis as a multitask learning problem, with tasks specific to age/sex groups.
• Extracted multiview features and utilized task-task and view-view relationships for enhanced learning.

Main Results:

• The Sparse-MVTC method demonstrated superior performance compared to existing state-of-the-art classification techniques.
• Experiments conducted on the ABIDE database validated the effectiveness of the proposed approach.
• The ensemble strategy integrated multiple classifiers for robust final diagnosis.

Conclusions:

• The proposed Sparse-MVTC ensemble learning method offers a significant advancement in computer-assisted ASD diagnosis.
• Accounting for age- and sex-related patterns in brain FC is crucial for accurate ASD detection.
• This approach holds promise for more precise and personalized diagnostic tools for ASD.