segcsvd PVS : A Convolutional Neural Network‐Based Tool for Quantification of Enlarged Perivascular Spaces ( PVS ) on T 1 ‐Weighted Images
Human Brain Mapping · 01.02.2026 · 47 (2)
DOI: 10.1002/hbm.70462 ↗Kurz gefasst
Fachartikel in der Zeitschrift Human Brain Mapping, online erschienen am 05.02.2026. Verfasst von Erin Gibson und 24 weiteren Autor:innen. Beteiligte Einrichtung: SB Centre for Brain Resilience & Recovery, Hurvitz Brain Sciences Program, Sunnybrook Research Institute, University of Toronto, Toronto, Ontario, Ca…. Hauptthemen laut MEDLINE: Convolutional Neural Networks, Magnetic Resonance Imaging, Glymphatic System, Cerebral Small Vessel Diseases. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: CIHR, Canadian Institutes of Health Research.
Abstract (Originalsprache)
Enlarged perivascular spaces (PVS) are imaging markers of cerebral small vessel disease (CSVD) that are associated with age, disease phenotypes, and overall health. Quantification of PVS is challenging but necessary to expand an understanding of their role in cerebrovascular pathology. Accurate and automated segmentation of PVS on T 1 ‐weighted images would be valuable given the widespread use of T 1 ‐weighted imaging protocols in multisite clinical and research datasets. We introduce segcsvd PVS , a convolutional neural network (CNN)‐based tool for automated PVS segmentation on T 1 ‐weighted images. segcsvd PVS was developed using a novel hierarchical approach that builds on existing tools and incorporates robust training strategies to enhance the accuracy and consistency of PVS segmentation. Performance was evaluated using a comprehensive evaluation strategy that included comparison to existing benchmark methods, ablation‐based validation, accuracy validation against manual ground truth annotations, correlation with age‐related PVS burden as a biological benchmark, and extensive robustness testing. segcsvd PVS achieved strong object‐level performance for basal ganglia PVS (DSC = 0.78), exhibiting both high sensitivity (SNS = 0.80) and precision (PRC = 0.78). Although voxel‐level precision was lower (PRC = 0.57), manual correction improved this by only ~3%, indicating that the additional voxels reflected primary boundary‐ or extent‐related differences rather than correctable false positive error. For non‐basal ganglia PVS, segcsvd PVS outperformed benchmark methods, exhibiting higher voxel‐level performance across several metrics (DSC = 0.60, SNS = 0.67, PRC = 0.57, NSD = 0.77), despite overall lower performance relative to basal ganglia PVS. Additionally, the association between age and segmentation‐derived measures of PVS burden was consistently stronger and more reliable for segcsvd PVS compared to benchmark methods across three cohorts (test6, ADNI, CAHHM), providing further evidence of the accuracy and consistency of its segmentation output. segcsvd PVS demonstrates robust performance across diverse imaging conditions and improved sensitivity to biologically meaningful associations, supporting its utility as a T 1 ‐based PVS segmentation tool.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- SB Centre for Brain Resilience & Recovery, Hurvitz Brain Sciences Program, Sunnybrook Research Institute, University of Toronto, Toronto, Ontario, Canada.
- Graduate Department of Psychological Clinical Science, University of Toronto Scarborough, Toronto, Ontario, Canada.
- University of Edinburgh, Edinburgh, UK.
- UK Dementia Research Institute Centre at the University of Edinburgh, Edinburgh, UK.
- University Health Network's Krembil Brain Institute, Edmond J Safra Program in Parkinson's Disease and the Rossy PSP Centre, Toronto Western Hospital, Toronto, Ontario, Canada.
- Division of Neurology, Toronto Western Hospital, University Health Network, University of Toronto, Toronto, Ontario, Canada.
ORCID-Profile: Erin Gibson, Maria Valdes Hernandez, Navneet Singh, Bradley J MacIntosh, Maged Goubran
Förderung
- CIHR
- Canadian Institutes of Health Research
Zitieren
APA
Gibson, E., Ramirez, J., Woods, L. A., Berberian, S., Ottoy, J., Scott, C. J. M., Yhap, V., Gao, F., Coello, R. D., Valdes Hernandez, M., Lang, A. E., Tartaglia, C. M., Kumar, S., Binns, M. A., Bartha, R., Symons, S., Swartz, R. H., Masellis, M., Singh, N., …, & Goubran, M. (2026). segcsvd PVS : A Convolutional Neural Network‐Based Tool for Quantification of Enlarged Perivascular Spaces ( PVS ) on T 1 ‐Weighted Images. Human Brain Mapping, 47(2). https://doi.org/10.1002/hbm.70462
Vancouver
Gibson E, Ramirez J, Woods LA, Berberian S, Ottoy J, Scott CJM, et al. segcsvd PVS : A Convolutional Neural Network‐Based Tool for Quantification of Enlarged Perivascular Spaces ( PVS ) on T 1 ‐Weighted Images. Human Brain Mapping. 2026;47(2). doi:10.1002/hbm.70462
BibTeX
@article{Gibson2026,
author = {Gibson, Erin and Ramirez, Joel and Woods, Lauren Abby and Berberian, Stephanie and Ottoy, Julie and Scott, Christopher J M and Yhap, Vanessa and Gao, Fuqiang and Coello, Roberto Duarte and Valdes Hernandez, Maria and Lang, Anthony E and Tartaglia, Carmela M and Kumar, Sanjeev and Binns, Malcolm A and Bartha, Robert and Symons, Sean and Swartz, Richard H and Masellis, Mario and Singh, Navneet and MacIntosh, Bradley J and Wardlaw, Joanna M and Black, Sandra E and ONDRI Investigators, ADNI, CAHHM Investigators, CAIN Investigators, colleagues from the Foundation Leducq Transatlantic Network of Excellence, ONDRI Investigators, ADNI, CAHHM Investigators, CAIN Investigators, colleagues from the Foundation Leducq Transatlantic Network of Excellence and Lim, Andrew S P and Goubran, Maged},
title = {segcsvd PVS : A Convolutional Neural Network‐Based Tool for Quantification of Enlarged Perivascular Spaces ( PVS ) on T 1 ‐Weighted Images},
journal = {Human Brain Mapping},
year = {2026},
volume = {47},
number = {2},
doi = {10.1002/hbm.70462}
}
EUVIMED