基于英国全国协作开放数据的二维超声心动图人工智能整体纵向应变测量
研究开发了基于机器学习的开放整体纵向应变(GLS)方法Unity-GLS,在6819张心尖四、二、三腔切面图像上训练,并用100例超声心动图的多专家共识进行外部验证。Unity-GLS与专家共识的误差为1.3 GLS单位,低于个体专家(2.00)、专有软件A(2.5)和B(2.2),相关性为0.91,高于个体专家(0.85)和专有软件(A 0.73,B 0.79)。
要点
- Unity-GLS与多专家共识的绝对误差为1.3 GLS单位,优于个体专家测量的2.00。
- Unity-GLS与专家共识的相关性为0.91,高于个体专家(0.85)和专有软件(A 0.73,B 0.79)。
- 以多专家视觉排序为参考,Unity-GLS的秩相关性为0.77,高于个体专家(0.72)和专有软件(A 0.70,B 0.74)。
核心结论
- 该开源GLS计算方法与专家共识的一致性达到个体专家测量和专有机器解决方案的水平,训练数据、代码和训练网络均可在网上免费获取。
研究设计与样本
- 研究训练多图像神经网络Unity-GLS,使用来自英国Unity UK超声心动图AI协作的6819张心尖四、二、三腔切面图像;外部验证数据集为100例超声心动图的上述三个切面图像,由11位专家分别标注收缩末期和舒张末期帧形成共识描记和点,并按纵向功能视觉分级排序;由一位专家使用两种专有软件包计算整体应变。
诊断标准与关键数值
- 以每例专家共识测量为参考标准,个体专家测量的中位绝对误差为2.00 GLS单位,Unity-GLS为1.3,专有软件A为2.5,专有软件B为2.2;与专家共识值的相关性分别为个体专家0.85、Unity-GLS 0.91、专有软件A 0.73、专有软件B 0.79;以多专家视觉排序为参考,个体专家应变测量的中位秩相关性为0.72,Unity-GLS为0.77,专有软件A为0.70,专有软件B为0.74;11位专家个体测量的中位GLS为-16.1(IQR:-19.3至-12.5)。
临床应用与阅片要点
- 作者开发的Unity-GLS开源方法可用于计算GLS,其与专家共识的一致性达到个体专家测量和专有机器解决方案的水平,训练数据、代码和训练网络均可在网上免费获取。
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: Global longitudinal strain (GLS) is reported to be more reproducible and prognostic than ejection fraction. Automated, transparent methods may increase trust and uptake.
OBJECTIVES: The authors developed open machine-learning-based GLS methodology and validate it using multiexpert consensus from the Unity UK Echocardiography AI Collaborative.
METHODS: We trained a multi-image neural network (Unity-GLS) to identify annulus, apex, and endocardial curve on 6,819 apical 4-, 2-, and 3-chamber images. The external validation dataset comprised those 3 views from 100 echocardiograms. End-systolic and -diastolic frames were each labelled by 11 experts to form consensus tracings and points. They also ordered the echocardiograms by visual grading of longitudinal function. One expert calculated global strain using 2 proprietary packages.
RESULTS: The median GLS, averaged across the 11 individual experts, was -16.1 (IQR: -19.3 to -12.5). Using each case's expert consensus measurement as the reference standard, individual expert measurements had a median absolute error of 2.00 GLS units. In comparison, the errors of the machine methods were: Unity-GLS 1.3, proprietary A 2.5, proprietary B 2.2. The correlations with the expert consensus values were for individual experts 0.85, Unity-GLS 0.91, proprietary A 0.73, proprietary B 0.79. Using the multiexpert visual ranking as the reference, individual expert strain measurements found a median rank correlation of 0.72, Unity-GLS 0.77, proprietary A 0.70, and proprietary B 0.74.
CONCLUSIONS: Our open-source approach to calculating GLS agrees with experts' consensus as strongly as the individual expert measurements and proprietary machine solutions. The training data, code, and trained networks are freely available online.
引用
- Stowell CC, Howard JP, Ng T, et al. 2-Dimensional Echocardiographic Global Longitudinal Strain With Artificial Intelligence Using Open Data From a UK-Wide Collaborative. JACC Cardiovasc Imaging. 2024. PMID: 39001730. doi:10.1016/j.jcmg.2024.04.017
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。