Deep Medial Voxels: Learned Medial Axis Approximations for Anatomical Shape Modeling
IEEE Transactions on Medical Imaging · 2025
DOI: 10.1109/tmi.2025.3582587 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Medical Imaging, 2025. Verfasst von Antonio Pepe und 5 weiteren Autor:innen. Hauptthemen laut MEDLINE: Deep Learning, Image Processing, Computer-Assisted, Imaging, Three-Dimensional, Models, Anatomic. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 69 Literaturangaben. Gefördert von: TU Graz LEAD Project Mechanics, Modeling and Simulation of Aortic Dissection, Alexander von Humboldt Foundation through German Federal Ministry of Education and Research and German Science Foundation Deutsche Forschungsgemeinschaft.
Abstract (Originalsprache)
Shape reconstruction from imaging volumes is a recurring need in medical image analysis. Common workflows start with a segmentation step, followed by careful post-processing and, finally, ad hoc meshing algorithms. As this sequence can be time-consuming, neural networks are trained to reconstruct shapes through template deformation. These networks deliver state-of-the-art results without manual intervention, but, so far, they have primarily been evaluated on anatomical shapes with little topological variety between individuals. In contrast, other works favor learning implicit shape models, which have multiple benefits for meshing and visualization. Our work follows this direction by introducing deep medial voxels, a semi-implicit representation that faithfully approximates the topological skeleton from imaging volumes and eventually leads to shape reconstruction via convolution surfaces. Our reconstruction technique shows potential for both visualization and computer simulations. Code available at https://github.com/apepe91/dmv.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Förderung
- TU Graz LEAD Project Mechanics, Modeling and Simulation of Aortic Dissection
- Alexander von Humboldt Foundation through German Federal Ministry of Education and Research and German Science Foundation Deutsche Forschungsgemeinschaft
Zitieren
APA
Pepe, A., Schussnig, R., Li, J., Gsaxner, C., Schmalstieg, D., & Egger, J. (2025). Deep Medial Voxels: Learned Medial Axis Approximations for Anatomical Shape Modeling. IEEE Transactions on Medical Imaging. https://doi.org/10.1109/tmi.2025.3582587
Vancouver
Pepe A, Schussnig R, Li J, Gsaxner C, Schmalstieg D, Egger J. Deep Medial Voxels: Learned Medial Axis Approximations for Anatomical Shape Modeling. IEEE Transactions on Medical Imaging. 2025. doi:10.1109/tmi.2025.3582587
BibTeX
@article{Pepe2025,
author = {Pepe, Antonio and Schussnig, Richard and Li, Jianning and Gsaxner, Christina and Schmalstieg, Dieter and Egger, Jan},
title = {Deep Medial Voxels: Learned Medial Axis Approximations for Anatomical Shape Modeling},
journal = {IEEE Transactions on Medical Imaging},
year = {2025},
doi = {10.1109/tmi.2025.3582587}
}
EUVIMED