Hierarchical Spatio-temporal Segmentation Network for Ejection Fraction Estimation in Echocardiography Videos

2025-08-26

AI Summary

Existing echocardiographic video segmentation methods achieve high accuracy in left ventricular endocardial delineation but suffer from accumulated local errors and inadequate long-range temporal modeling, leading to substantial ejection fraction (EF) estimation bias and limited clinical reliability. To address these limitations, we propose a hierarchical spatiotemporal segmentation framework that jointly preserves local anatomical details and captures global cardiac dynamics. Specifically, we introduce a spatiotemporal cross-scanning module to concurrently model intra-frame anatomy and inter-frame ventricular motion, and innovatively integrate convolutional networks with the Mamba architecture to enhance long-horizon temporal dependency modeling while reducing computational complexity. Evaluated on multiple clinical datasets, our method significantly improves EF estimation accuracy—reducing mean absolute error by 32.7%—and outperforms state-of-the-art segmentation-based EF estimation approaches. This work provides a more robust and interpretable solution for automated cardiac functional assessment.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Mixture of Experts (MoE)
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Automated segmentation of the left ventricular endocardium in echocardiography videos is a key research area in cardiology. It aims to provide accurate assessment of cardiac structure and function through Ejection Fraction (EF) estimation. Although existing studies have achieved good segmentation performance, their results do not perform well in EF estimation. In this paper, we propose a Hierarchical Spatio-temporal Segmentation Network (ourmodel) for echocardiography video, aiming to improve EF estimation accuracy by synergizing local detail modeling with global dynamic perception. The network employs a hierarchical design, with low-level stages using convolutional networks to process single-frame images and preserve details, while high-level stages utilize the Mamba architecture to capture spatio-temporal relationships. The hierarchical design balances single-frame and multi-frame processing, avoiding issues such as local error accumulation when relying solely on single frames or neglecting details when using only multi-frame data. To overcome local spatio-temporal limitations, we propose the Spatio-temporal Cross Scan (STCS) module, which integrates long-range context through skip scanning across frames and positions. This approach helps mitigate EF calculation biases caused by ultrasound image noise and other factors.

Problem

• Improving ejection fraction estimation accuracy in echocardiography videos
• Automated left ventricular endocardium segmentation with spatio-temporal modeling
• Mitigating EF calculation biases from ultrasound noise and limitations

Innovation

• Hierarchical network with convolutional and Mamba stages
• Spatio-temporal Cross Scan module for long-range context
• Balances single-frame details with multi-frame dynamics