Abstract
Accurate delineation of mitral valve anatomy across complementary imaging modalities is essential for quantitative assessment, intervention planning, and intraoperative guidance. This paper presents a unified submission to the MVAA 2026 challenge covering CT, three dimensional transesophageal echocardiography (3D TEE), and surgical video segmentation. The CT task combines STU-Net with a confidence masked nnU-Net through probability fusion and conservative connected component filtering. The 3D TEE task integrates a standard nnU-Net with four residual encoder nnU-Net models using weighted softmax ensembling and label specific morphological refinement. The surgical video task employs U-Net++ with an EfficientNet-B4 encoder, followed by component aware filtering and auxiliary model consistency checks. Model selection jointly considers overlap and surface distance criteria, and all inference procedures are deterministic and compatible with a unified Docker environment. On the reported validation sets, the CT ensemble achieved Dice 0.8573, ASD 0.2286mm, HD 4.1553mm, and HD95 0.8674~mm. The balanced 3D TEE ensemble achieved DSC 0.8561, HD 10.4295, and ASD 0.5773. For surgical video, the final method achieved DSC 0.8008 and reduced HD from 167.50 to 72.26 relative to the unrefined primary prediction. These results demonstrate that modality specific modeling, probability based complementarity, and conservative structural refinement provide an effective and reproducible solution for multimodal mitral valve segmentation.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MVAA_027.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=JB28e6aOSq
BibTex
@InProceedings{ZhaJia_Mitral_MICCAISAT2026,
author = { Zhang, Jiayi AND Xu, Han AND He, Ruihan AND Qiu, Xihe AND Gu, Lei AND Eils, Roland},
title = { { Mitral Valve Segmentation in CT, 3D TEE, and Surgical Video } },
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}
}