Accepted Articles
Back to the articles list |
Back to browse issues page
Department of Biomedical Engineering, Faculty of Electrical Engineering, K. N. Toosi University of Technology, Tehran, Iran
Abstract:
Introduction: Diffusion MRI combined with deterministic tractography enables the reconstruction of whole-brain structural networks. However, the inherent noise in measurements and probabilistic nature of fiber tracking generate an uncertain number of false white matter connections. Current limitations in network-level anatomical data make it difficult to reliably separate real connections from artifactual ones. While network thresholding methods are frequently used to filter out presumably spurious connections, their varying effects on network characteristics and subsequent statistical analyses remain almost unclear.
Method: We analyzed data from 27 schizophrenia patients and 27 demographically matched healthy controls. Five network weighting schemes (fiber density, streamline count, mean fiber length, apparent diffusion coefficient, and global fractional anisotropy) were examined under two systematic thresholding approaches (absolute and proportional) across multiple threshold levels. Network properties were quantified using three standard metrics: node degree, clustering coefficient, and global efficiency. Group comparisons were performed using independent samples t-tests.
Results: We found that lower threshold values tended to yield more significant differences in graph metrics compared to higher thresholds. Additionally, proportional thresholding produced more consistent patterns of metric reduction across all weighting methods. Among the different weightings, fiber density exhibited the greatest statistical differences between patients and healthy controls.
Conclusion: Our findings demonstrate that the choice of threshold significantly impacts graph metrics and statistical outcomes, potentially influencing study conclusions. These results highlight the need for more rigorous justification in selecting thresholding methods and suggest that researchers should consider adopting multiple analytical approaches to ensure robustness in network-based analyses.
Method: We analyzed data from 27 schizophrenia patients and 27 demographically matched healthy controls. Five network weighting schemes (fiber density, streamline count, mean fiber length, apparent diffusion coefficient, and global fractional anisotropy) were examined under two systematic thresholding approaches (absolute and proportional) across multiple threshold levels. Network properties were quantified using three standard metrics: node degree, clustering coefficient, and global efficiency. Group comparisons were performed using independent samples t-tests.
Results: We found that lower threshold values tended to yield more significant differences in graph metrics compared to higher thresholds. Additionally, proportional thresholding produced more consistent patterns of metric reduction across all weighting methods. Among the different weightings, fiber density exhibited the greatest statistical differences between patients and healthy controls.
Conclusion: Our findings demonstrate that the choice of threshold significantly impacts graph metrics and statistical outcomes, potentially influencing study conclusions. These results highlight the need for more rigorous justification in selecting thresholding methods and suggest that researchers should consider adopting multiple analytical approaches to ensure robustness in network-based analyses.
Type of Study: Original |
Subject:
Computational Neuroscience
Received: 2025/08/13 | Accepted: 2025/12/23
Received: 2025/08/13 | Accepted: 2025/12/23
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
Copyright © The Author(s);
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC-By-NC), which permits use, distribution, and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
Contact Information
