Investigating topological alterations in procedural memory network across neuropsychiatric disorders using rs-fMRI and graph theory

Abstract

Background: Cognitive network dysfunction represents a core pathophysiological feature across major neuropsychiatric disorders, including Attention Deficit Hyperactivity Disorder (ADHD), bipolar disorder (BD), and schizophrenia (SZ). The procedural memory network (PMN), involving cortico-striatal-cerebellar circuits, is vital for skill learning and automatic cognition. However, its topological changes and link to cognitive impairments have not been studied across major neuropsychiatric disorders.

Methods: This study analyzed resting-state functional MRI (rs-fMRI) data from 40 individuals with ADHD, 49 with BD, 50 with SZ, and 50 healthy controls (HCs). PMN was defined using 34 regions of interest (ROIs) from Harvard-Oxford Atlas, with graph theory measures calculated for all regions. Significant network disruptions emerged, showing altered local efficiency (LE), average path length (APL), and degree (P < 0.05) across groups.

Results: Key findings show that in ADHD, increased APL in left cerebellar lobule VII indicates disrupted information flow and emotional processing, while decreased connectivity in the right claustrum may impair integration and working memory. In BD, reduced LE in right cerebellar lobule II is linked to attention and motor control deficits; increased APL in lobules I and VIII suggests disrupted network communication and emotional processing; and decreased connectivity in the right subthalamic nucleus and lobule VIII may contribute to mood swings and attention problems. In SZ, decreased LE in right putamen and left cerebellar lobule VIII relates to working memory and emotional processing deficits; reduced APL in right caudate and cerebellar lobule II implies more effort for regional communication; and increased connectivity in the caudate and right cerebellar lobules I and II likely reflects compensatory or pathological hyperactivity. Comparisons indicate SZ shows increased connectivity in the claustrum and cerebellar lobule I, unlike ADHD which shows decreases in these areas; SZ has lower network efficiency but higher caudate connectivity than BD, which has more cerebellar and subthalamic disruptions; and BD shows decreased connectivity in the claustrum and subthalamic nucleus compared to ADHD, which has more cerebellar and attention network changes.

Conclusion: These findings suggest that the PMN, particularly its segregation and integration properties, plays a key role in explaining cognitive deficits in ADHD, BD, and SZ.

Clinical trial number: Not applicable.

Supplementary Information: The online version contains supplementary material available at 10.1186/s12868-025-00979-z.

Keywords: Brain mapping; Cognitive dysfunction; Functional connectivity; Graph theory; Procedural memory; Resting-state fMRI.

Fig. 1

Schematic representation of the PMN construction and graph theoretical analysis pipeline

Fig. 2

(a, b, and c) 3D views of coronal, sagittal, and axial planes of the brain’s PMN regions (l: left, r: right)

Fig. 3

Regions of the PMN showing significant differences in graph theory measures (average path length and degree) between patients with ADHD and HCs. The size of each brain ROI reflects the mean value (± standard deviation) of the measure across patients

Fig. 4

Regions of the PMN showing significant differences in graph theory measures (average path length, local efficiency and degree) between patients with BD and HCs. The size of each brain ROI reflects the mean value (± standard deviation) of the measure across patients

Fig. 5

Regions of the PMN showing significant differences in graph theory measures (average path length, local efficiency and degree) between patients with SZ and HCs. The size of each brain ROI reflects the mean value (± standard deviation) of the measure across patients

Fig. 6

Regions of the PMN showing significant differences in graph theory measures (average path length, local efficiency and degree) between patients with SZ and ADHD. The size of each brain ROI reflects the mean value (± standard deviation) of the measure across patients

Fig. 7

Regions of the PMN showing significant differences in graph theory measures (average path length and degree) between patients with SZ and BD. The size of each brain ROI reflects the mean value (± standard deviation) of the measure across patients

Fig. 8

Regions of the PMN showing significant differences in graph theory measures (local efficiency and degree) between patients with ADHD and BD. The size of each brain ROI reflects the mean value (± standard deviation) of the measure across patients