中国医学装备2026，Vol.23Issue(2)：13-17,5.DOI:10.3969/j.issn.1672-8270.2026.02.003
深度学习重建算法在低剂量冠状动脉CT血管成像中的图像质量与辐射剂量优化:一项前瞻性随机对照研究
Optimization of image quality and radiation dose of low dose coronary CTA by using deep learning reconstruction algorithms:a prospective RCT
摘要
Abstract
Objective:To compare the differences of coronary computed tomography angiography(CCTA)images among low tube current combined with deep learning image reconstruction(DLIR)algorithm,low tube current combined with hybrid iterative reconstruction(HIR)algorithm,and conventional scanning protocol combined with HIR algorithm,so as to assess clinical application value of DLIR algorithm in CCTA.Methods:A total of 100 patients who scheduled for retrospective electrocardiogram(ECG)-gated CCTA examinations at The Affiliated Yongchuan Hospital of Chongqing Medical University from August 2023 to May 2024 were prospectively enrolled.Patients were randomly divided into a conventional dose group(n=50)and a low dose group(n=50)using random number table.The conventional dose group underwent scanning at 160mAs,and image reconstruction of that was conducted by using an iterative reconstruction algorithm(Karl 3D).The low dose group underwent scanning at 60mAs,and the scanning data of this group were reconstructed respectively by two kinds of algorithms,which were further divided into group A and group B(group A used Karl 3D,and group B used DLIR).Radiation dose,evaluation of subjective image quality,and measured values of objective image quality,which included image noise(SD),signal-to-noise ratio(SNR),and contrast-to-noise ratio(CNR),were compared among conventional dose group,low lose group A and low dose group B.Results:The effective radiation dose[(4.29±0.90)mSv]in the low dose group(including group A and B)was significantly lower than[(9.38±1.90)mSv]in the conventional dose group(t=17.10,P<0.05).There was not statistically significant difference in subjective evaluation of image quality between the group B and the conventional dose group(P>0.05).The image quality of conventional dose group and group B were better than that of group A,and the differences were significant(x2=39.71,46.22,P<0.05).The background noise[(12.37±2.54)HU]of the group B was significantly lower than[(23.98±4.93)HU]of the conventional dose group,and[(28.70±5.41)HU]of the group A of the low dose group,and the differences were statistically significant(t=14.80,-19.32,P<0.05),respectively.There were not significant differences in CT values at aortic root and various segments of the coronary artery among the conventional dose group,group A and group B(P>0.05).There was significant difference in noise among three groups(F=176.39,P<0.05).In addition,the differences of SNR and CNR of various segments[right coronary artery(RCA),left anterior descending(LAD)and left circumflex artery(LCX)]of the coronary artery among three groups were significant(FSNR=132.79,129.36,133.37,FCNR=161.23,170.68,169.64,P<0.05).Conclusions:The application of low tube current combined with the DLIR algorithm in retrospective ECG-gated CCTA can significantly reduce radiation dose,and further improve image quality.关键词
冠状动脉CT血管成像(CCTA)/深度学习重建(DLIR)算法/混合迭代重建(HIR)/辐射剂量Key words
Coronary computed tomography angiography(CCTA)/Deep learning image reconstruction(DLIR)algorithm/Hybrid iterative reconstruction(HIR)/Radiation dose分类
医药卫生引用本文复制引用
杨欢,周永霞,丁昭军,刘文罡..深度学习重建算法在低剂量冠状动脉CT血管成像中的图像质量与辐射剂量优化:一项前瞻性随机对照研究[J].中国医学装备,2026,23(2):13-17,5.基金项目
重庆市科卫联合医学科研项目(2024MSXM112) Chongqing Science and Health joint Medical Research Project(2024MSXM112) （2024MSXM112）
