fNIRS-Based Minimum Spanning Tree Analysis in Neuropsychiatric Disorders

Background: Functional near-infrared spectroscopy (fNIRS) is a neuroimaging method for evaluating cortical network dynamics during cognitive tasks. Conventional graph pipelines require threshold/density choices that can bias between-subject comparisons. We therefore use the minimum spanning tree (MST)—a unique, cycle-free tree connecting all recorded regions at minimal cost without thresholding—to obtain a standardized, density-invariant network representation comparable across participants and cognitive conditions to study the neuro-psychiatric problems as Obsessive Compulsive Disorder (OCD), Schizophrenia (SCHIZO) and Migraine . Objective: This study examined prefrontal cortex network topology differences across patients groups via MST analysis computed from fNIRS data during a Stroop task, aiming to detect diagnosis-related changes, to relate network organization to behavior, and to assess hemispheric similarity and connectivity. Methods: 16-channel fNIRS recordings were obtained under neutral, congruent, and incongruent conditions. Partial correlation--based functional connectivity matrices were transformed into MSTs to compute seven global metrics, survival ratio, and hemispheric edge counts; reaction time and accuracy were also measured. Results: Schizophrenia (SCHIZO) showed a less integrated network than controls, with longer characteristic path length and trends toward larger diameter, lower leaf fraction, and lower degree divergence. Slower responses correlated with less integrated MST patterns. Within-group similarity was lower in OCD and SCHIZO, while hemispheric organization was stable. Migraine showed no robust differences from controls. Conclusion: MST metrics offer unbiased, threshold-free descriptors of large-scale brain organization, revealing diagnosis-related topology changes in task-based fNIRS studies.