Japanese Team Maps ADHD Brain Networks
A new preprint from Japan's Computational Psychiatry team decomposed resting-state fMRI dynamics, revealing unique network alterations in schizophrenia, ASD, and ADHD. The research uses advanced brain imaging to identify distinct neural signatures for each condition. This builds on recent JAMA Psychiatry work that identified three ADHD biotypes through morphometric similarity networks.
The RIKEN Computational Brain Dynamics Team in Japan, led by Okito Yamashita, focuses on creating novel, imaging-based diagnostics by analyzing human brain data from fMRI, MEG, and EEG scans. Their work on decomposing resting-state fMRI utilizes methods like Dynamic Mode Decomposition (DMD) to extract and interpret complex spatiotemporal patterns from brain activity. The JAMA Psychiatry study identified three ADHD biotypes by analyzing brain structure. One biotype is linked to severe overall symptoms and emotional dysregulation with alterations in the medial prefrontal cortex. A second shows hyperactivity/impulsivity with changes in the anterior cingulate cortex, while the third is primarily inattentive with alterations in the superior frontal gyrus. Distinguishing between ADHD, Autism Spectrum Disorder (ASD), and schizophrenia is a major challenge as they can have overlapping symptoms and genetic risk factors. Meta-analyses of fMRI studies confirm that while the conditions share some altered brain activation patterns, more prominent disorder-specific neural signatures have been identified for both ASD and ADHD. Despite these research advances, no primary psychiatric disorder can be definitively diagnosed through neuroimaging alone. Brain scans like fMRI are currently used in clinical psychiatry primarily to rule out other medical conditions, while their application for diagnosing psychiatric illnesses remains a research tool to identify group-level differences. This type of research reflects a broader shift in psychiatry away from rigid diagnostic categories. Scientists are increasingly focusing on "transdiagnostic" and dimensional approaches, which examine biological markers and symptom severity across different conditions to create more precise treatments. The technique of Dynamic Mode Decomposition (DMD) is borrowed from fields like fluid dynamics and is particularly well-suited for analyzing time-series data. It allows researchers to extract dynamic patterns from resting-state fMRI data, providing information about how brain networks evolve and interact over time.
