CLAS

心脏功能智能定量——从超声序列精准估计射血分数等心功能指标

射血分数（Ejection Fraction, EF）是评估心功能的核心指标，临床依赖专家逐帧描记，主观性强、耗时。 CLAS（Cardiac function quantification via Co-Learning of Appearance and Shape）提出外观-形态 协同学习框架：

  • 分割与回归联合优化，直接从序列级图像估计 EF
  • 时序一致性约束（Temporal-consistent Segmentation），提升运动边界稳定性
  • 与心腔分割端到端协同，结果可回溯、可解释

(Wei et al., 2023; Wei et al., 2020; Xue et al., 2021)

🚀 在线演示 （上传心尖切面序列，自动输出 EF 及容积曲线）

References

2023

  1. Co-learning of appearance and shape for precise ejection fraction estimation from echocardiographic sequences
    Hongrong Wei, Junqiang Ma, Yongjin Zhou, and 2 more authors
    IF 10.9
    Medical Image Analysis, 2023
    DOI

2021

  1. Left ventricle quantification challenge: A comprehensive comparison and evaluation of segmentation and regression for mid-ventricular short-axis cardiac MR data
    Wufeng Xue, Jiahui Li, Zhiqiang Hu, and 8 more authors
    IF 6.7
    IEEE journal of biomedical and health informatics, 2021
    DOI

2020

  1. Temporal-consistent Segmentation of Echocardiography with Co-learning from Appearance and Shape
    Hongrong Wei, Heng Cao, Yiqin Cao, and 4 more authors
    CCF-B
    In Medical Image Computing and Computer Assisted Intervention (MICCAI 2020), 2020
    DOI