冠状动脉计算机断层血管造影智能分析在冠状动脉慢性完全闭塞病变中应用的研究

The feasibility study of intelligent analysis system of CCTA in chronic total occlusion of coronary artery

目的:通评估基于深度学习(DL)算法的冠状动脉CTA(CCTA)智能分析技术在辅助诊断冠状动脉慢性完全闭塞病变(CTO)中的可行性。方法:基于卷积神经网络自动分割、重建CCTA图像,同时提供辅助诊断信息。回顾性分析我院2020年1月1日至2020年12月31日期间,行CCTA检查的冠状动脉CTO患者100例,所有入组患者同时接受了数字减影血管造影检查(DSA)。以DSA检查为参考标准,比较基于DL的CCTA智能分析技术与人工判读诊断CTO病变的准确性。结果:与人工判读相比,CCTA智能分析技术显著缩短了约80%的后处理和诊断时间,分别仅需(2.7±0.9)min和(1.4±0.6)min,其在冠状动脉中度狭窄中的诊断准确率与人工判读相近(89.3%vs.89.8%,P>0.05);而智能分析技术对于重度狭窄与CTO病变的诊断准确性不及传统人工方法(重度狭窄:59.9%vs.74.3%;完全闭塞:61.3%vs.80.8%;P<0.05)。结论:基于DL的CCTA智能分析技术可提高冠状动脉粥样硬化病变的评估效率,在冠心病患者中度狭窄病变中的诊断准确性较高,但其在重度狭窄与CTO病变中的诊断鉴别能力还有待提高。

Objective:To investigate the feasibility of deep learning(DL)-based intelligent analysis system of CCTA in diagnosing the chronic total occlusion(CTO)of coronary artery.Methods:Convolutional neural network was used to automatically segment and reconstruct CCTA images of patients with coronary atherosclerosis and provide auxiliary diagnostic information.Retrospective analysis was performed on 100 patients with coronary CTO lesions who underwent CCTA examination in our hospital from January 1,2020,to December 31,2020.All enrolled patients also received digital subtraction angiography(DSA).DSA was used as the reference standard to compare the accuracy of the intelligent coronary CTA analysis system and the traditional interpretation method in diagnosing CTO lesions.Results:Compared with traditional interpretationmethods,the DL-based CCTA intelligent analysis system significantly reduces the post-processing and diagnosis time by about 80%,only(2.7±0.9)min and(1.4±0.6)min,respectively.The diagnostic accuracy of intelligent CCTA analysis system in middle coronary artery stenosis lesions was similar to that of traditional interpretation(89.3%vs.89.8%,P>0.05),but the diagnostic accuracy of severe stenosis and CTO lesions was inferior to that of traditional interpretation(severe stenosis:59.9%vs.74.3%;CTO:61.3%vs.80.8%;all P<0.05).Conclusions:DL-based intelligent coronary artery analysis system has greatly improved the evaluation efficiency of coronary atherosclerotic lesions,with high diagnostic accuracy in moderate coronary artery stenosis lesions,but its diagnostic differentiation ability in severe stenosis and CTO lesions remains to be further improved.