Li Yidan, Liang Lirong, Guo Dichen, Yang Yuanhua, Gong Juanni, Zhang Xinyuan, Zhang Di, Jiang Zhe, Lu Xiuzhang
Department of Echocardiography, Heart Center, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China.
Clinical Epidemiology & Tobacco Dependence Treatment Research Department, Beijing Institute of Respiratory Medicine, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China.
Front Med (Lausanne). 2021 Aug 23;8:697396. doi: 10.3389/fmed.2021.697396. eCollection 2021.
Right ventricular (RV) function plays a vital role in the prognosis of patients with chronic thromboembolic pulmonary hypertension (CTEPH). We used new machine learning (ML)-based fully automated software to quantify RV function using three-dimensional echocardiography (3DE) to predict adverse clinical outcomes in CTEPH patients. A total of 151 consecutive CTEPH patients were registered in this prospective study between April 2015 and July 2019. New ML-based methods were used for data management, and quantitative analysis of RV volume and ejection fraction (RVEF) was performed offline. RV structural and functional parameters were recorded using 3DE. CTEPH was diagnosed using right heart catheterization, and 62 patients underwent cardiac magnetic resonance to assess right heart function. Adverse clinical outcomes were defined as PH-related hospitalization with hemoptysis or increased RV failure, including conditions requiring balloon pulmonary angioplasty or pulmonary endarterectomy, as well as death. The median follow-up time was 19.7 months (interquartile range, 0.5-54 months). Among the 151 CTEPH patients, 72 experienced adverse clinical outcomes. Multivariate Cox proportional-hazard analysis showed that ML-based 3DE analysis of RVEF was a predictor of adverse clinical outcomes (hazard ratio, 1.576; 95% confidence interval (CI), 1.046~2.372; = 0.030). The new ML-based 3DE algorithm is a promising technique for rapid 3D quantification of RV function in CTEPH patients.
右心室（RV）功能在慢性血栓栓塞性肺动脉高压（CTEPH）患者的预后中起着至关重要的作用。我们使用基于机器学习（ML）的新型全自动软件，通过三维超声心动图（3DE）对RV功能进行量化，以预测CTEPH患者的不良临床结局。在2015年4月至2019年7月期间，共有151例连续的CTEPH患者纳入了这项前瞻性研究。采用基于ML的新方法进行数据管理，并离线对RV容积和射血分数（RVEF）进行定量分析。使用3DE记录RV结构和功能参数。通过右心导管检查诊断CTEPH，62例患者接受心脏磁共振成像以评估右心功能。不良临床结局定义为与肺动脉高压相关的咯血住院或RV衰竭加重，包括需要球囊肺动脉血管成形术或肺动脉内膜剥脱术的情况以及死亡。中位随访时间为19.7个月（四分位间距，0.5 - 54个月）。在151例CTEPH患者中，72例出现了不良临床结局。多变量Cox比例风险分析显示，基于ML的3DE分析RVEF是不良临床结局的一个预测指标（风险比，1.576；95%置信区间（CI），1.046~2.372； = 0.030）。基于ML的新3DE算法是一种很有前景的技术，可用于快速三维量化CTEPH患者的RV功能。
