Functional connectivity differences in adult's ADHD - a MEG study

The neurobiology of adult Attention-Deficit/Hyperactivity Disorder (ADHD), particularly functional brain network connectivity, remain poorly understood. Magnetoencephalography (MEG) can reveal frequency-specific network dynamics given its high temporal resolution. Here, we investigated intrinsic functional connectivity differences between adults with ADHD (n = 24) and healthy controls (n = 44) using MEG data from the Open MEG Archive (OMEGA) dataset. We employed source reconstruction (Destrieux atlas), weighted phase-lag index (wPLI) connectivity across six frequency bands, graph theory metrics (Characteristic Path Length (CPL), node strength, clustering coefficient), and network-based statistics (NBS). We observed widespread hypo-connectivity in the high-gamma band (50-150 Hz) in adults with ADHD compared to controls. NBS identified a significant high-gamma sub-network, predominantly involving dorsal and ventral attention networks (DAN/VAN) and default mode network (DMN) nodes centered around a left fusiform gyrus hub, where all constituent connections exhibited consistently lower connectivity in the ADHD group. Globally, this was reflected in reduced gamma network integration (longer CPL) within the DAN and VAN. Locally, reduced high-gamma clustering was observed in VAN nodes (e.g., insula) and reduced node strength in a DAN region (postcentral sulcus). Predictive modeling using ElasticNet regression confirmed the importance of high-gamma metrics, with CPL measures yielding moderate classification accuracy (AUC {approx} 0.70-0.73). In contrast to high-gamma findings, the alpha band showed increased integration (shorter CPL) within the DMN and between the VAN and DAN in the ADHD group, alongside differences in alpha and beta band node properties in cingulate and somatomotor regions. Our findings reveal robust, frequency-dependent functional network alterations in adult ADHD, particularly highlighting disrupted high-frequency communication within and between key cognitive networks.