Abstract
Mitral valve assessment is essential for the diagnosis, treatment planning, and intraoperative guidance of structural heart disease. Automated analysis across heterogeneous imaging modalities remains challenging because cardiac CT, 3D transesophageal echocardiography (TEE), and intraoperative frames differ substantially in spatial resolution, intensity distribution, anatomical visibility, and annotation format. We develop a reproducible baseline framework for three MVAA tasks: pre-procedural CT segmentation, intraoperative 3D TEE segmentation, and intraoperative frame segmentation. CT and 3D TEE volumes are converted into nnU-Net v2 datasets and trained with default 3D full-resolution models, residual-encoder variants, and boundary-aware losses. CT segmentation is further extended with pseudo-label training on unlabeled volumes while preserving the held-out validation split. Intraoperative frames are handled by a DINOv3 UPerNet model targeting the evaluated anatomical label. On internal validation, the best CT model achieves Dice 0.865, Hausdorff distance (HD) 4.97, and average surface distance (ASD) 0.245. The best 3D TEE model achieves Dice 0.842, HD 10.70, and ASD 0.592. The best intraoperative-frame model achieves Dice 0.830, HD 86.97, and ASD 14.04. These results indicate that modality-specific preprocessing, accurate target-label handling, and boundary-aware optimization are important for robust multi-modal mitral valve anatomy analysis.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MVAA_014.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=POEJUdBr54
BibTex
@InProceedings{YanZi_BoundaryAware_MICCAISAT2026,
author = { Yang, Zi AND Wu, Zhigang AND Ding, Xueqian AND Wang, Xiaojuan AND Fu, Yu},
title = { { Boundary-Aware and Semi-Supervised Baselines for Multi-Modal Mitral Valve Anatomy Analysis } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17268},
month = {pending},
page = {pending}
}