Multimodal brain network topology and enhanced computer-aided diagnosis in Parkinson’s Disease: a systematic review and meta-analysis - Report - MDSpire
Advertisement
Multimodal brain network topology and enhanced computer-aided diagnosis in Parkinson’s Disease: a systematic review and meta-analysis
Brain Network Alterations and Diagnostic Accuracy in Parkinson’s Disease
Overview
This systematic review and meta-analysis of 80 studies involving 3736 Parkinson’s disease (PD) patients identified distinct patterns of brain network alterations across neuroimaging modalities. Diffusion MRI revealed deficits in both network segregation and integration, functional MRI showed mainly reduced segregation, while structural MRI and EEG showed no consistent abnormalities. The findings support the potential of graph theoretical analysis metrics combined with computational methods for improving early PD diagnosis.
Background
Parkinson’s disease is a prevalent neurodegenerative disorder with increasing socioeconomic impact. Diagnosis currently relies on clinical assessment, which is challenging especially in atypical or early stages. Noninvasive neuroimaging techniques enable characterization of brain connectivity, which can be quantified using graph theoretical analysis (GTA) to assess the brain's global network topology. PD is conceptualized as a disconnection syndrome, but prior studies have reported inconsistent findings regarding brain network alterations. This meta-analysis aimed to clarify consistent multimodal network changes and evaluate the diagnostic utility of GTA metrics.
Data Highlights
Modality
Metric
Effect Size (g)
P-value
Direction in PD vs HC
dMRI
Clustering Coefficient
-0.328
0.002
Lower
dMRI
Local Efficiency
-0.272
0.007
Lower
dMRI
Global Efficiency
-0.445
<0.001
Lower
dMRI
Characteristic Path Length
0.396
0.001
Higher
dMRI
Normalized Clustering Coefficient
0.245
0.026
Higher
fMRI
Clustering Coefficient
-0.351
0.004
Lower
fMRI
Local Efficiency
-0.217
0.066
Trend Lower
fMRI
Global Efficiency
NS
0.844
No Change
fMRI
Characteristic Path Length
NS
0.996
No Change
fMRI
Modularity
0.217
0.036
Higher
sMRI
Various Metrics
NS
NS
No Consistent Abnormalities
EEG
Various Metrics
NS
NS
No Consistent Abnormalities
Key Findings
Diffusion MRI demonstrated significant reductions in network segregation (clustering coefficient, local efficiency) and integration (global efficiency), with increased characteristic path length in PD patients compared to healthy controls.
Functional MRI showed mainly decreased network segregation (lower clustering coefficient and increased modularity) but no significant changes in network integration metrics.
Structural MRI and EEG studies did not reveal consistent global network topology abnormalities in PD.
The multimodal analysis indicates modality-specific patterns of brain network disruption rather than a single convergent alteration across imaging types.
Graph theoretical metrics hold promise as biomarkers for early and accurate diagnosis of PD when combined with computational diagnostic techniques.
Clinical Implications
These findings highlight the utility of diffusion and functional MRI-based graph theoretical metrics as potential biomarkers for Parkinson’s disease, particularly for early detection and differentiation from healthy aging. Clinicians and researchers should consider multimodal neuroimaging approaches to capture distinct aspects of brain network pathology in PD. Integration of these metrics into computer-assisted diagnostic tools may enhance diagnostic accuracy and facilitate timely intervention.
Conclusion
This comprehensive meta-analysis confirms consistent brain network alterations in Parkinson’s disease, predominantly detected by diffusion and functional MRI. The results support the development of graph theory-based biomarkers combined with computational methods to improve early diagnosis and disease management.
References
Comprehensive Analysis of Brain Network Structures and Improved Computer-Assisted Diagnosis in Parkinson’s Disease: A Systematic Review and Meta-Analysis
