Abstract
Mitral valve anatomy analysis is a clinically important yet technically challenging problem because the valve is thin, deformable, and observed through heterogeneous imaging modalities. The MVAA 2026 challenge evaluates automated mitral valve segmentation across cardiac computed tomography (CT), three-dimensional transesophageal echocardiography (3D TEE), and surgical video. This paper describes the baseline methods released for the challenge. The baseline intentionally combines reliable, reproducible components rather than highly specialized task-specific engineering: 3D U-Net models are used for volumetric CT and TEE segmentation, while a two-dimensional U-Net++ model with an EfficientNet encoder is used for video-frame segmentation. To reflect the limited annotation regime, the CT and surgical-video branches additionally use exponential-moving-average teacher models and confidence-filtered pseudo-label supervision on unlabeled data. All predictions are evaluated by the unified MVAA validation protocol using Dice similarity coefficient (DSC), Hausdorff distance (HD), and average surface distance (ASD). On the local validation split, the baseline obtains DSC, HD, and ASD scores of 0.6184, 9.6213, and 0.9594 for CT; 0.7451, 18.7673, and 1.6552 for 3D TEE; and 0.6440, 193.2962, and 24.7622 for surgical video. These results provide a transparent reference point for future multimodal and geometry-aware methods.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MVAA_035.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=0f3qbdlqhg
BibTex
@InProceedings{DenBo_Baseline_MICCAISAT2026,
author = { Deng, Bo AND Bai, Jieyun},
title = { { Baseline Method of the MVAA 2026 Challenge for Multimodal Mitral Valve Segmentation } },
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}
}