MTCNet: Motion and Topology Consistency Guided Learning for Mitral Valve Segmentationin 4D Ultrasound

2025-07-01

AI Summary

Addressing challenges in mitral valve segmentation from 4D ultrasound—including severe motion artifacts, poor image quality, scarce annotated data, and inadequate modeling of inter-phase dependencies—this paper proposes a semi-supervised cross-phase consistency learning framework. Its key contributions are: (1) motion-guided consistency constraints that leverage temporal motion priors to enhance inter-phase feature alignment; (2) topology-aware correlation regularization that incorporates anatomical structure priors to ensure morphologically plausible segmentations; and (3) a bidirectional attention memory bank enabling efficient spatiotemporal feature propagation across phases. Evaluated on a large-scale 4D ultrasound dataset comprising 160 patients and 1,408 cardiac phases, the method achieves a Dice score of 87.30% and a Hausdorff distance of 1.75 mm—outperforming state-of-the-art approaches, particularly in cross-phase segmentation consistency and robustness under motion corruption.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multimodal Learning
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Mitral regurgitation is one of the most prevalent cardiac disorders. Four-dimensional (4D) ultrasound has emerged as the primary imaging modality for assessing dynamic valvular morphology. However, 4D mitral valve (MV) analysis remains challenging due to limited phase annotations, severe motion artifacts, and poor imaging quality. Yet, the absence of inter-phase dependency in existing methods hinders 4D MV analysis. To bridge this gap, we propose a Motion-Topology guided consistency network (MTCNet) for accurate 4D MV ultrasound segmentation in semi-supervised learning (SSL). MTCNet requires only sparse end-diastolic and end-systolic annotations. First, we design a cross-phase motion-guided consistency learning strategy, utilizing a bi-directional attention memory bank to propagate spatio-temporal features. This enables MTCNet to achieve excellent performance both per- and inter-phase. Second, we devise a novel topology-guided correlation regularization that explores physical prior knowledge to maintain anatomically plausible. Therefore, MTCNet can effectively leverage structural correspondence between labeled and unlabeled phases. Extensive evaluations on the first largest 4D MV dataset, with 1408 phases from 160 patients, show that MTCNet performs superior cross-phase consistency compared to other advanced methods (Dice: 87.30%, HD: 1.75mm). Both the code and the dataset are available at https://github.com/crs524/MTCNet.

Problem

• Accurate 4D mitral valve segmentation in ultrasound
• Overcoming limited phase annotations and motion artifacts
• Ensuring anatomically plausible inter-phase consistency

Innovation

• Motion-guided consistency learning with bi-directional attention
• Topology-guided correlation regularization for anatomical plausibility
• Semi-supervised learning with sparse phase annotations