Exploiting the Brain's Network Structure for Automatic Identification of ADHD Subjects.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Medical Imaging

Background:

• Attention Deficit Hyperactive Disorder (ADHD) is a prevalent neurodevelopmental disorder impacting children.
• Resting-state functional Magnetic Resonance Imaging (fMRI) offers insights into brain function and network alterations in ADHD.

Approach:

• Functional brain networks were modeled using voxel-wise activity correlations from fMRI data.
• A Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA) classifier was trained on network features.
• A novel masking approach identified and utilized region-specific features to enhance classification accuracy.

Key Points:

• Graph-motif features, specifically 3-cycle maps, effectively capture network properties relevant to ADHD.
• A masking strategy focusing on discriminative brain regions significantly improved classification performance.
• The study achieved a classification accuracy of 69.59% on the ADHD-200 challenge dataset.

Conclusions:

• Automatic classification of ADHD using resting-state fMRI is feasible.
• Network-based analysis combined with targeted feature selection offers a promising avenue for ADHD diagnosis and understanding.
• The proposed method demonstrates potential for clinical application in identifying ADHD subjects.