Deep learning image reconstruction can reduce the radiation dose and contrast agent volume in patients undergoing coronary computed tomography angiography
CHEN Chang
BIAN Chuanzhen
MEI Junqing
MA Hongbing
Abstract：Objective To validate the feasibility of the deep learning image reconstruction(DLIR)algorithm in coronary computed tomography angiography(CCTA)under low radiation dose and low contrast agent volume conditions.Methods This prospective study included 86 patients with normal BMI who underwent CCTA at the Affiliated BenQ Hospital of Nanjing Medical University from November 2021 to April 2022.The patients were randomly divided into group A and group B.Both groups employed Smart-mA tube current automatic control technology,Auto Gating,Smart Phase and Motion correction algorithm techniques,with a noise index set at 12.2 HU.Iodixanol(350 mgI/mL)was used as the contrast agent.The tube voltage was set to 70 kV,with the contrast agent volume calculated as(body weight×0.275)mL in group A,and the tube voltage was set to 120 kV,the contrast agent volume was(body weight×0.55)mL in group B.Group A used the DLIR algorithm for image reconstruction,while group B used the 50%ASIR-V algorithm.The CT values and noise levels of the aortic root,left main,left anterior descending artery,left circumflex artery,and right coronary artery proximal segments were measured and calculated.Objective evaluation parameters,including signal-to-noise ratio(SNR),contrast-to-noise ratio(CNR),and edge rise distance were computed.A double-blind method was used to compare the subjective image quality of the two reconstruction methods.Results Except for no significant differences in CNR of the left main artery and SNR of the left circumflex artery proximal segment(P=0.358,0.252),the CNR and SNR of all other regions of interest in group A were significantly higher than those in group B(P<0.001).The edge rise distance of the left anterior descending proximal segment in group A was smaller than in group B(P<0.001).Image quality in both groups met diagnostic requirements,but group A demonstrated significantly better subjective image quality than group B(P<0.001).Radiation dose:The effective dose in group A was 0.81±0.40 mSv,compared to 2.84±1.50 mSv in group B,with a statistically significant difference(P<0.001).Contrast agent volume:The volume in group A was 22.11±3.31 mL,while in group B it was 34.40±2.98 mL,with a statistically significant difference(P<0.001).Conclusion The DLIR algorithm can effectively reduce radiation dose and contrast agent volume in CCTA,demonstrating potential for wider application.
Keywords：deep learning image reconstructionradiation dosecoronary computed tomography angiography
Publication Date：2025-01-27
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 76-81 )