Towards reproducible perivascular space quantification: an open-source segmentation benchmark
Poster No:
1820
Submission Type:
Abstract Submission
Authors:
Merel van der Thiel1, Eva van Heese2, Britt van den Heuvel1, Valery Molina3, Maria Jaramillo3, Jacobus Jansen1, Brady Williamson4, Jose Bernal5
Institutions:
1Maastricht University Medical Center, Maastricht, Netherlands, 2Amsterdam University Medical Center, Amsterdam, Netherlands, 3Universidad del Valle, Cali, Colombia, 4Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, 5German Centre for Neurodegenerative Diseases (DZNE), Magdeburg, Germany
First Author:
Co-Author(s):
Introduction:
The quantification of perivascular spaces (PVS) -a promising non-invasive marker of cerebral waste clearance [1]- in structural magnetic resonance imaging (MRI) has skyrocketed in recent years [2]. Unlike traditional visual ratings [3], computational methods can reduce subjectivity and laboriousness, minimize floor and ceiling effects, and provide deeper insights into PVS morphology (e.g., volume, length, width, and tortuosity) [4]. A generalized lack of publicly available code and benchmarks leads to duplicated efforts and hinders large-scale research on PVS. The absence of standardized pipelines further impedes reproducibility. In response, we created an open-source repository to compile PVS quantification tools and validate them against a shared ground truth.
Methods:
Repository:
We established a PVS quantification Github repository (https://github.com/PVS-segmentation-repository), providing guidelines for code submissions and showcasing current contributions from the community (Fig.1).
Approach:
Phase 1) Code collection: Instructions for contributing PVS quantification code are now available on the Github page. To ensure compatibility and streamline the validation and testing of contributions, we will assist in containerizing submissions. Therefore, there are no requirements for programming language or algorithm type.
Phase 2) Method comparison: We will compare submitted quantification tools against each other and against a manually segmented ground truth dataset. An in-depth code review will assess user-friendliness and functionality. Ultimately, we aim to develop guidelines for selecting the appropriate quantification pipeline based on input data characteristics.
Dataset: The reference dataset includes T1-weighted, FLAIR and T2-weighted (two echo times) images from 40 healthy elderly subjects scanned on ultra-high field MRI (7T Magnetom, Siemens Healthcare) in Maastricht, the Netherlands. PVS masks were manually delineated by two independent raters, serving as the designated 'ground truth' (Fig.2).
We established a PVS quantification Github repository (https://github.com/PVS-segmentation-repository), providing guidelines for code submissions and showcasing current contributions from the community (Fig.1).
Approach:
Phase 1) Code collection: Instructions for contributing PVS quantification code are now available on the Github page. To ensure compatibility and streamline the validation and testing of contributions, we will assist in containerizing submissions. Therefore, there are no requirements for programming language or algorithm type.
Phase 2) Method comparison: We will compare submitted quantification tools against each other and against a manually segmented ground truth dataset. An in-depth code review will assess user-friendliness and functionality. Ultimately, we aim to develop guidelines for selecting the appropriate quantification pipeline based on input data characteristics.
Dataset: The reference dataset includes T1-weighted, FLAIR and T2-weighted (two echo times) images from 40 healthy elderly subjects scanned on ultra-high field MRI (7T Magnetom, Siemens Healthcare) in Maastricht, the Netherlands. PVS masks were manually delineated by two independent raters, serving as the designated 'ground truth' (Fig.2).
·Fig 1. Visual representation highlighting the motivation and aims of the open-source perivascular space (PVS) segmentation code repository.
·Fig 2. Example images and acquisition parameters of the 7 Tesla reference dataset including T1-weighted, FLAIR and T2-weighted images with manual perivascular space (PVS) segmentations (red).
Results:
Call for code:
We encourage scientists developing PVS quantification algorithms to share their code in our open-source repository. By sharing their code on the platform, researchers will gain visibility for their work, contribute to community-driven standards, and benefit from collaborative improvements and comparative validation against other quantification methods. To demonstrate public presentation, we have included a vanilla Frangi filter-based approach in our repository. Additionally, we have contacted a subset of corresponding authors (n=25) from publications featuring PVS quantification methods to invite them to contribute to our cause, and several researchers have already expressed interest.
How to contribute:
We welcome all types of PVS quantification code, including preprocessing, processing, and postprocessing tools. There will be a continuous call for code contributions and we strongly encourage researchers to share their PVS quantification code. For submission details or how to join the PVS quantification segmentation repository team, please visit: https://github.com/PVS-segmentation-repository.
We encourage scientists developing PVS quantification algorithms to share their code in our open-source repository. By sharing their code on the platform, researchers will gain visibility for their work, contribute to community-driven standards, and benefit from collaborative improvements and comparative validation against other quantification methods. To demonstrate public presentation, we have included a vanilla Frangi filter-based approach in our repository. Additionally, we have contacted a subset of corresponding authors (n=25) from publications featuring PVS quantification methods to invite them to contribute to our cause, and several researchers have already expressed interest.
How to contribute:
We welcome all types of PVS quantification code, including preprocessing, processing, and postprocessing tools. There will be a continuous call for code contributions and we strongly encourage researchers to share their PVS quantification code. For submission details or how to join the PVS quantification segmentation repository team, please visit: https://github.com/PVS-segmentation-repository.
Conclusions:
The project's current goal is to gather PVS quantification code from various groups, such that code snippets which otherwise would have remained private become open to the research community. Contributions beyond segmentation, such as preprocessing, visualization, or post-processing tools, would also be highly beneficial to the community. Long-term, we aim to establish a common open-source benchmark repository for PVS segmentation to improve transparency and reproducibility in future PVS-related research studies. The PVS repository team emphasizes the need for transparency and standardization, highlighting that a common codebase can greatly enhance reproducibility and elevate PVS-related studies.
Modeling and Analysis Methods:
Image Registration and Computational Anatomy
Segmentation and Parcellation 2
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Anatomy and Functional Systems
Neuroinformatics and Data Sharing:
Databasing and Data Sharing 1
Workflows
Keywords:
Cerebro Spinal Fluid (CSF)
Cerebrovascular Disease
HIGH FIELD MR
Machine Learning
Open Data
Open-Source Code
Open-Source Software
Segmentation
STRUCTURAL MRI
Other - perivascular space; neurofluids
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Provide references using APA citation style.
2. Waymont, J. M. J. et al. Systematic review and meta-analysis of automated methods for quantifying enlarged perivascular spaces in the brain. Neuroimage 297, 120685 (2024).
3. Potter, G. M. et al. Enlarged perivascular spaces and cerebral small vessel disease. Int. J. Stroke 10, 376–381 (2015).
4. Wardlaw, J. M. et al. Perivascular spaces in the brain: anatomy, physiology and pathology. Nat. Rev. Neurol. 16, 137–153 (2020).