Diminished small-world properties and compensatory reorganization of resting-state functional brain networks in patients with subacute spontaneous intracerebral hemorrhage: a graph-theoretic analysis of topological properties and their relationship with cognitive impairment

Cognitive impairment is common after intracerebral hemorrhage (ICH), yet conventional imaging metrics fail to explain its neural mechanisms from the perspective of whole-brain functional networks. This study aims to investigate abnormal changes in small-world topological properties of resting-state functional brain networks in patients with spontaneous ICH during the subacute phase, to characterize the topological reorganization at both global and nodal levels, and to further elucidate the relationship between differential topological metrics and cognitive impairment as well as the influence of lesion volume.

An observational cross-sectional study was conducted. A total of 28 patients with subacute spontaneous ICH (ICH group) who were treated at Department of Neurosurgery of Huizhou Central People's Hospital between December 2024 and January 2026, and 35 healthy controls (HC) matched for age, sex, and years of education who were treated at Health Check-up Center or General Clinic of Huizhou Central People's Hospital during the same period were enrolled. All participants underwent Mini-Mental State Examination (MMSE) assessment and 3. 0 T resting-state functional MRI (rs-fMRI) scanning. For ICH patients, T1 structural images were subjected to semi-automatic lesion delineation using ITK-SNAP software, and mirror normalization was performed based on the Clinical Toolbox for SPM to reduce interference from focal lesions on spatial normalization. rs-fMRI data were preprocessed using the DPABI software package, including head motion correction, tissue segmentation, spatial normalization, linear detrending, nuisance regression, and bandpass filtering. When extracting the mean BOLD time series from 116 brain regions of interest defined by the automated anatomical labeling (AAL) atlas, voxels falling within the lesion and hematoma masks were excluded. Functional connectivity matrices were constructed using Pearson correlation coefficients followed by Fisher r-to-z transformation. The GRETNA toolbox was used to construct binary undirected functional brain networks within a sparsity threshold range of 0. 05 to 0. 40, and the area under the curve (AUC) was calculated for global topological metrics including clustering coefficient (Cp), characteristic path length (Lp), normalized clustering coefficient (Gamma), normalized characteristic path length (Lambda), small-world property (Sigma), global efficiency (Eglob), and local efficiency (Eloc), as well as nodal metrics including nodal efficiency (Ne) and nodal local efficiency (NLe).

Both groups exhibited small-world topological characteristics (Lambda≈1, Sigma>1); however, the ICH group demonstrated significant decreases in Sigma, Gamma, Eloc, Eglob, and Lambda (all FDR-corrected P<0. 05), an increase in Lp (FDR-corrected P=0. 001), and no statistical difference in Cp (FDR-corrected P=0. 086). At the nodal level, following Bonferroni correction, the ICH group exhibited decreased nodal efficiency in the left globus pallidus, bilateral amygdala, right posterior cingulate gyrus, left putamen, left cerebellar Crus Ⅰ, and cerebellar vermis (P<0. 05), and increased nodal efficiency in the left medial superior frontal gyrus and left supplementary motor area (P<0. 05). Regarding nodal local efficiency, the left globus pallidus, left amygdala, and right posterior cingulate gyrus were decreased (P<0. 05). Correlation analysis revealed that nodal efficiency of the left medial superior frontal gyrus was negatively correlated with spatial orientation scores (r=-0. 452, P=0. 023), nodal efficiency of the left amygdala was negatively correlated with tongue twister repetition scores (r=-0. 433, P=0. 031), and nodal local efficiency of the right posterior cingulate gyrus was negatively correlated with tongue twister repetition scores (r=-0. 422, P=0. 036). No significant correlations were observed between global topological metrics and cognitive scores. Lesion volume was negatively correlated only with spatial orientation scores (r=-0. 572, P=0. 016), and showed no significant correlation with any of the functional network topological metrics.

Patients with subacute ICH exhibit whole-brain functional network reorganization characterized by diminished small-world properties, decreased integration efficiency, and impaired local information processing, suggesting that cognitive impairment after ICH has a broad network basis.