机器学习在冠状动脉计算机断层扫描领域的应用及进展

Application and progress of machine learning in coronary computed tomography angiography

心脏计算机断层扫描血管造影术(CCTA)已成为评估冠状动脉疾病的重要非侵入性手段,随着其在临床的广泛应用及图像分析特征的增加,CCTA图像评估对技术及时间的要求不断提高。机器学习(ML)是人工智能的分支领域,它完全由数据驱动,通过计算机算法对大型数据集中变量的潜在关系进行识别及分析,实现对外部数据的预测。在心脏CT领域,不同ML算法的应用可提高CCTA的成像效率及质量,有助于病变评估及风险分层,同时也为心脏功能学成像提供了新的应用。该文主要对ML在心脏CT图像分析、风险模型、CT心肌灌注及CT血流储备分数中的应用研究进展进行综述。

Cardiac computed tomography angiography(CCTA)has become an important non-invasive method to evaluate coronary artery disease.With the extensive application and increased image analysis features,more demands on operational technique and efficiency are asked.Machine learning(ML)is the subarea of artificial intelligence(AI),which is completely data driven,by computer algorithm to identify and analyze the potential relationship of centralized variables in large data sets for realizing the prediction of external data.In the field of cardiac CT,the application of various ML algorithms would improve the efficiency and quality of CCTA,helping accurate lesion assessment and risk stratification.It also brings new applications in cardiac functional imaging.The applications of ML in cardiac CT have been reviewed in present paper including CT-image analysis,risk stratification,CT-myocardial perfusion and CT-fractional flow reserve.