Characterization of Human Midbrain Vasculature from High-Resolution Post-Mortem MRI Data
Poster No:
1759
Submission Type:
Abstract Submission
Authors:
Marshall Xu1, Kevin Sitek2, Omer Faruk Gulban3, Saskia Bollmann4
Institutions:
1The University of Queesland, Brisbane, Queensland, 2Northwestern University, Evanston, IL, 3Maastricht University, Maastricht, Netherlands, 4The University of Queensland, Brisbane, Queensland
First Author:
Co-Author(s):
Introduction:
The midbrain is a critical hub for motor control, sensory processing and autonomic regulation, relying on a well-organized vascular network to support oxygen delivery and metabolic demands. Disruptions in the vasculature can impact brain function and lead to neurodegenerative conditions (Hirsch et al., 2012). Recent work has characterized the geometric features of the dorsal midbrain (including superior and inferior colliculus) from a post-mortem subject (Sitek et al., 2024). In this study, we segmented the vasculature of the whole post-mortem midbrain specimen to extract geometric statistics and better understand its structural organization. Furthermore, we conducted a graph-based analysis of the whole midbrain vascular segmentation to explore its vascular connectivity.
Methods:
Data acquisition: High-resolution T2*-weighted anatomical MRI was collected with 50 μm isotropic resolution from a post-mortem human brainstem (65-year-old male, no neurological conditions) using a small-bore 7 Tesla MRI scanner (Calabrese et al., 2015).
Vessel segmentation: To avoid memory overflow, we truncated the specimen into 6 slabs in the axial direction. the brainstem was divided into six axial slabs. Each slab underwent segmentation using a morphology-based small vessel segmentation pipeline (OMELETTE) (Mattern, 2021), which generated initial segmentations. This pipeline produced some false negatives (e.g., missing small vessels) and false positives (e.g., thickened vascular boundaries and artifacts near the aqueduct). To refine the segmentation, we employed the Booster module of VesselBoost (Xu et al., 2024) with a 3D version of UNet3+ (Huang et al., 2020) as the base model architecture. Finally, segmentations of all slabs were stitched to create a continuous 3D representation of the midbrain vasculature.
Graph extraction: We used the Voreen (Volume Rendering Engine) (Meyer-Spradow et al., 2009) framework to analyse the geometric features (number of segments, mean length of segments, mean radius per segment and mean segment tortuosity) of the whole midbrain vasculature, as well as the graph-based analysis. The vessel graph was generated via a scalable iterative pipeline (Drees et al., 2021). To avoid any noise and large 'chunky' artefacts, we excluded the largest 0.1% of the segments and any segments with length or diameter less than 50 μm (the image resolution) for statistical analysis.
Vessel segmentation: To avoid memory overflow, we truncated the specimen into 6 slabs in the axial direction. the brainstem was divided into six axial slabs. Each slab underwent segmentation using a morphology-based small vessel segmentation pipeline (OMELETTE) (Mattern, 2021), which generated initial segmentations. This pipeline produced some false negatives (e.g., missing small vessels) and false positives (e.g., thickened vascular boundaries and artifacts near the aqueduct). To refine the segmentation, we employed the Booster module of VesselBoost (Xu et al., 2024) with a 3D version of UNet3+ (Huang et al., 2020) as the base model architecture. Finally, segmentations of all slabs were stitched to create a continuous 3D representation of the midbrain vasculature.
Graph extraction: We used the Voreen (Volume Rendering Engine) (Meyer-Spradow et al., 2009) framework to analyse the geometric features (number of segments, mean length of segments, mean radius per segment and mean segment tortuosity) of the whole midbrain vasculature, as well as the graph-based analysis. The vessel graph was generated via a scalable iterative pipeline (Drees et al., 2021). To avoid any noise and large 'chunky' artefacts, we excluded the largest 0.1% of the segments and any segments with length or diameter less than 50 μm (the image resolution) for statistical analysis.
Results:
Before conducting statistical analysis on the stitched segmentation, we manually refined some contours of the midbrain to minimize errors (Figure 1). Based on the data generated by Voreen, we identified a total of 173,124 segments corresponding to vasculature in the T2*-weighted midbrain specimen. These segments had a mean length of 459.9 μm (range: 50.0–4473.8 µm) and a mean radius of 95.4 μm (range: 27.9–263.2 µm). Approximately 50% of the segments had lengths between 150 μm and 450 μm, and over 60% had radii between 50 μm and 100 μm.
In terms of curvature and connectivity, the statistical characteristics mirrored those of length and radius. Both exhibited right-skewed distributions, with values concentrated towards smaller lengths and radii (Figure 2).
In terms of curvature and connectivity, the statistical characteristics mirrored those of length and radius. Both exhibited right-skewed distributions, with values concentrated towards smaller lengths and radii (Figure 2).
·Figure 1 (A) The MinIPs of the post-mortem midbrain sample. The generated segmentation shown as red labels. (B) Extracted vessel graph of the dorsal region (C) 3D reconstruction of the segmentation
·Figure 2 Vessel length, radius, curveness and node of degree statistical plots.
Conclusions:
Our work establishes an initial framework for exploring human midbrain vascular architecture by integrating a deep-learning-based segmentation pipeline with graph-based analysis. Future directions include extending this framework to in vivo brainstem vessel segmentation to investigate subcortical hemodynamics and the vascular contributions to brain dysfunction. Additionally, refining the segmentation to capture smaller vessels and enhancing vascular continuity will be crucial for improving the accuracy of graph-based analyses.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
Methods Development
Segmentation and Parcellation 2
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Subcortical Structures 1
Keywords:
Brainstem
Segmentation
Sub-Cortical
Other - Post mortem
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.
Drees, D., Scherzinger, A., Hägerling, R., Kiefer, F., & Jiang, X. (2021). Scalable robust graph and feature extraction for arbitrary vessel networks in large volumetric datasets. BMC Bioinformatics, 22(1), 346. https://doi.org/10.1186/s12859-021-04262-w
Hirsch, S., Reichold, J., Schneider, M., Székely, G., & Weber, B. (2012). Topology and Hemodynamics of the Cortical Cerebrovascular System. Journal of Cerebral Blood Flow & Metabolism, 32(6), 952–967. https://doi.org/10.1038/jcbfm.2012.39
Huang, H., Lin, L., Tong, R., Hu, H., Zhang, Q., Iwamoto, Y., Han, X., Chen, Y.-W., & Wu, J. (2020). UNet 3+: A Full-Scale Connected UNet for Medical Image Segmentation. ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 1055–1059. https://doi.org/10.1109/ICASSP40776.2020.9053405
Mattern, H. (2021). Openly available sMall vEsseL sEgmenTaTion pipelinE (OMELETTE). Proc. Intl. Soc. Mag. Reson. Med. ISMRM.
Meyer-Spradow, J., Ropinski, T., Mensmann, J., & Hinrichs, K. (2009). Voreen: A Rapid-Prototyping Environment for Ray-Casting-Based Volume Visualizations. IEEE Computer Graphics and Applications, 29(6), 6–13. https://doi.org/10.1109/MCG.2009.130
Sitek, K., Xu, M., Gulban, O. F., & Bollmann, S. (2024). Segmentation and quantification of mesoscopic subcortical vessels using post mortem MRI at 50 micron. OHBM, Seoul.
Xu, M., Ribeiro, F. L., Barth, M., Bernier, M., Bollmann, S., Chatterjee, S., Cognolato, F., Gulban, O. F., Itkyal, V., Liu, S., Mattern, H., Polimeni, J. R., Shaw, T. B., Speck, O., & Bollmann, S. (2024). VesselBoost: A Python Toolbox for Small Blood Vessel Segmentation in Human Magnetic Resonance Angiography Data. Aperture Neuro, 4. https://doi.org/10.52294/001c.123217
Yushkevich, P. A., Piven, J., Hazlett, H. C., Smith, R. G., Ho, S., Gee, J. C., & Gerig, G. (2006). User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. NeuroImage, 31(3), 1116–1128. https://doi.org/10.1016/j.neuroimage.2006.01.015