Abstract
Accurate vessel segmentation is essential for vascular disease assessment, yet automatic methods often produce fragmented and incomplete vessels due to low vessel-to-background contrast in non-contrast CT (NCCT) and curved vascular morphology in contrast-enhanced CT (CECT). To address this issue, we propose a geometry-aware 3D Vessel Completion Network (VessComNet) that restores missing vessel segments in both NCCT and CECT. The framework leverages reliable CECT annotations to train a unified cross-protocol model, enabling vessel completion in challenging NCCT images without manual NCCT annotations. By incorporating centerline-guided geometric priors, the method preserves anatomical connectivity and structural plausibility. Experiments on three datasets demonstrate that our approach effectively restores missing regions and improves vessel continuity.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ShapeMI_015.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=PnvwhmqP6j
BibTex
@InProceedings{YaoLin_VessComNet_MICCAISAT2026,
author = { Yao, Linlin AND Batheja, Vivek AND Balamuralikrishna, Praveen Thoppey Srinivasan AND Pickhardt, Perry J. AND Grayson, Peter C. AND Summers, Ronald M.},
title = { { VessComNet: Geometry-Aware 3D Vessel Completion Across Non-Contrast and Contrast-Enhanced CT Images } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17259},
month = {pending},
page = {pending}
}
