Abstract:Attention deficit hyperactivity disorder (ADHD) is a highly prevalent psychiatric disorder, which is generally characterized by symptoms of inattentiveness, hyperactivity and impulsivity. As a popular functional imaging technique, the resting-state functional magnetic resonance imaging (rsfMRI) is frequently used to explore the neural mechanism of ADHD. However, after resting-state brain networks are obtained from the rsfMRI dataset with the traditional independent component analysis, it is difficult to detect reliable ADHD-related brain networks via voxel-level inference because of the high dimensions and relatively few samples of fMRI data. To address the problem, a novel framework and index using the dual temporal and spatial sparse representation (DTSSR) is proposed, which is able to detect the resting-state brain networks related to ADHD from a large scale network-level perspective in an rsfMRI dataset consisting of 22 adult ADHD patients. First, the group-wise brain networks and coupling parameters are inferred from the ADHD fMRI dataset via DTSSR. These coupling parameters are then processed by average-pooling as an index for activity of each group-wise network. Finally, the Spearman correlation analysis between those obtained indices and the ADHD rating scores is performed to detect the networks related to ADHD. Experiment demonstrates that the dorsal attention network and executive control network are significantly related to ADHD. The result shows high stability under different dictionary sizes and can be reasonably explained from brain science. The proposed framework can be used to explore the underlying neural mechanism of ADHD in a new perspective.

Detecting and Analyzing Brain Networks of Adult ADHD Using Dual Temporal and Spatial Sparse Representation