Characterization of cortical morphology and structural covariance network features in post-stroke patients with visuospatial neglect
10.3969/j.issn.1672-5921.2025.07.004
- VernacularTitle:卒中后视空间忽略患者皮质形态与皮质结构协变网络的特征分析
- Author:
Wanying ZHAO
1
;
Lei CAO
1
;
Weiqun SONG
1
;
Linlin YE
1
Author Information
1. 100053 北京,首都医科大学宣武医院康复医学科
- Publication Type:Journal Article
- Keywords:
Stroke;
Visuospatial neglect;
Cortical morphology;
Structural covariance network
- From:
Chinese Journal of Cerebrovascular Diseases
2025;22(7):474-486
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate cortical morphological changes and structural covariance network(SCN)topological features in patients with visuospatial neglect(VSN)after stroke.Methods A retrospective analysis was conducted on acute stroke patients admitted consecutively to the Department of Rehabilitation Medicine,Xuanwu Hospital,Capital Medical University,from December 2023 to February 2025.General and clinical data were collected and compared,including age,sex,stroke type(hemorrhagic or ischemic),and time from stroke onset to first brain MRI.All patients were assessed for VSN using the line bisection task,line cancellation test,and star cancellation test.Patients with VSN were assigned to the VSN group,and those without VSN to the non-VSN group.Brain MRI was used to collect patients' brain structural images.T1-weighted MRI data were processed using Freesurfer for whole-brain segmentation.Based on the Desikan-Killiany atlas,the cortex was parcellated into 68 regions per hemisphere,total intracranial volume and cortical parameters were extracted,including cortical area(CA),mean cortical curvature(CC),cortical thickness(CT),and cortical volume(CV).A region-specific asymmetry index(AI)was calculated for cortical parameters of each brain regions using(left-right)/(left+right)to assess cortical lateralization.Analyze the SCN of each cortical parameters through covariance matrix to reflect covariation pattern of each brain region's structural morphological alternations.SCN were constructed separately for each cortical parameter based on inter-regional Pearson correlation coefficients(absolute values),with intracranial volume,age,and sex controlled via linear regression.The resulting 68×68 SCN matrices were transformed into binary networks across a sparsity range of 0.1 to 0.4(in 0.01 steps),and graph theoretical analysis were performed.Permutation tests were used to compare the global and local graph theoretical metrics of cortical SCN under each cortical parameters with different sparsity.Global metrics included clustering coefficient,path length,small-worldness parameters(normalized clustering coefficient[Gamma],normalized path length[Lambda],and small-world index[Sigma]),global efficiency,and average local efficiency.Local metrics included nodal degree,betweenness centrality,and nodal efficiency.Results(1)A total of 109 acute stroke patients were included(81 males,28 females,aged 30-80 years,mean[64±9]years),with 54 in the VSN group and 55 in the non-VSN group.No significant differences were found in age,sex,stroke type,or time from stroke onset to MRI between two groups(all P＞0.05).(2)Comparison of cortical features(of different brain regions)between two groups showed significantly larger CA in the right insular surface area in the VSN group(2 466.50[2 143.75,2 662.50]mm2 vs.2 128.00[1 961.00,2 479.00)]mm2,P=0.037).No significant differences were observed in other regions or metrics(all P＞0.05).The VSN group has significantly lower AI values for CT in the right isthmus cingulate gyrus(-0.01[-0.05,0.03]vs.0.02[-0.01,0.06],P=0.028)and postcentral gyrus(-0.02[-0.04,0.00]vs.0.00[-0.02,0.02],P=0.026).No significant AI differences were observed in other regions or metrics(all P＞0.05).(3)For global network metrics:in the CA-based VSN,compared with the non-VSN group,Sigma values were significantly higher in the VSN group across multiple sparsity thresholds and at the average sparsity level(P＜0.01).In the CT-based SCN,Lambda values were significantly higher in the VSN group at most sparsity thresholds and at the average level(P=0.004).In the CC-based SCN,clustering coefficient and average local efficiency were significantly lower in the VSN group across most sparsity thresholds and average sparsity(P＜0.01).No significant differences were found in global metrics within the CV-based SCN.(4)For local metrics:in the CV-based SCN,compared with the non-VSN group,the VSN group showed significantly lower nodal efficiency in the left inferior temporal gyrus at most sparsity thresholds and at the average level(0.35[0.29,0.50]vs.0.65[0.51,0.72],P＜0.01).No significant differences in local metrics were observed in other SCN between the two groups.Conclusions Patients with VSN exhibit abnormal cortical morphology and SCN topology,characterized by reduced overall integration efficiency and weakened local connectivity,alongside enhanced small-worldness and possible compensatory reorganization.The conclusions of this study require further validation in multicenter,large-scale,prospective studies.
