深度学习导出的心肌应变
本研究开发了开源深度学习应变(DLS)分析流程,从标准超声心动图B型图像回顾性测量整体纵向应变(GLS)。在外部验证中,DLS与二维GLS呈中等一致性(ICC 0.56),偏差-3.31%;在晚期心脏淀粉样变性患者中ICC为0.64。DLS的变异低于人工测量,且快速、不依赖厂商。
要点
- DLS与二维GLS在外部验证中呈中等一致性(ICC 0.56),偏差-3.31%(一致性界限-11.65%至5.02%)。
- 传统GLS的厂商间组内相关系数为0.29,低于操作者间0.63,提示厂商间变异较大。
- 在晚期心脏淀粉样变性患者中,DLS与二维GLS的ICC为0.64,偏差0.57%,一致性界限-4.87%至6.01%。
核心结论
- 开源DLS提供的变异低于人工测量,定量结果相似;该方法快速、一致、不依赖厂商、已公开,并适用于广泛的图像质量范围。
研究设计与样本
- 本研究开发并验证自动化深度学习应变(DLS)分析流程;DLS基于EchoNet-Dynamic的语义分割结果,通过计算左心室心内膜轮廓长度变化导出纵向应变;在大型外部数据集及多种导致心脏肥厚的疾病人群中评估。
诊断标准与关键数值
- 传统GLS的厂商间组内相关系数为0.29(95%CI:-0.01至0.53,P=0.03);操作者间组内相关系数为0.63(95%CI:0.48-0.74,P<0.001)。
- 初始输入轮廓微小变化时,逐步像素偏移导致平均绝对误差3.48%,6像素偏移对应应变比例差异13.52%。
- 外部验证中,DLS与二维GLS的组内相关系数(ICC)为0.56(P=0.002),偏差-3.31%(一致性界限:-11.65%至5.02%)。
- 在晚期心脏淀粉样变性患者中,DLS与二维GLS的ICC为0.64(95%CI:0.53-0.72,P<0.001),偏差0.57%,一致性界限-4.87%至6.01%。
- DLS方法在心脏肥厚人群间显示出差异(P<0.0001)。
临床应用与阅片要点
- DLS可从标准超声心动图B型图像回顾性测量GLS,不依赖厂商,适用于图像质量范围广。
- DLS与二维GLS呈中等一致性,在晚期心脏淀粉样变性患者中偏差较小(0.57%)。
- 传统GLS厂商间变异较大(ICC 0.29),提示跨厂商比较时需注意。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
BACKGROUND: Echocardiographic strain measurements require extensive operator experience and have significant intervendor variability. Creating an automated, open-source, vendor-agnostic method to retrospectively measure global longitudinal strain (GLS) from standard echocardiography B-mode images would greatly improve post hoc research applications and may streamline patient analyses.
OBJECTIVES: This study was seeking to develop an automated deep learning strain (DLS) analysis pipeline and validate its performance across multiple applications and populations.
METHODS: Interobserver/-vendor variation of traditional GLS, and simulated effects of variation in contour on speckle-tracking measurements were assessed. The DLS pipeline was designed to take semantic segmentation results from EchoNet-Dynamic and derive longitudinal strain by calculating change in the length of the left ventricular endocardial contour. DLS was evaluated for agreement with GLS on a large external dataset and applied across a range of conditions that result in cardiac hypertrophy.
RESULTS: In patients scanned by 2 sonographers using 2 vendors, GLS had an intraclass correlation of 0.29 (95% CI: -0.01 to 0.53, P = 0.03) between vendor measurements and 0.63 (95% CI: 0.48-0.74, P < 0.001) between sonographers. With minor changes in initial input contour, step-wise pixel shifts resulted in a mean absolute error of 3.48% and proportional strain difference of 13.52% by a 6-pixel shift. In external validation, DLS maintained moderate agreement with 2-dimensional GLS (intraclass correlation coefficient [ICC]: 0.56, P = 0.002) with a bias of -3.31% (limits of agreement: -11.65% to 5.02%). The DLS method showed differences (P < 0.0001) between populations with cardiac hypertrophy and had moderate agreement in a patient population of advanced cardiac amyloidosis: ICC was 0.64 (95% CI: 0.53-0.72), P < 0.001, with a bias of 0.57%, limits of agreement of -4.87% to 6.01% vs 2-dimensional GLS.
CONCLUSIONS: The open-source DLS provides lower variation than human measurements and similar quantitative results. The method is rapid, consistent, vendor-agnostic, publicly released, and applicable across a wide range of imaging qualities.
引用
- Kwan AC, Chang EW, Jain I, et al. Deep Learning-Derived Myocardial Strain. JACC Cardiovasc Imaging. 2024. PMID: 38551533. doi:10.1016/j.jcmg.2024.01.011
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。