Feasibility Study of Triple-low CCTA for Coronary Artery Disease Screening Combining Contrast Enhancement Boost and Deep Learning Reconstruction.

Background: The aim of this study was to compare the image quality of coronary computed tomography angiography (CCTA) images obtained using contrast enhancement boost (CE-boost) technology combined with deep learning reconstruction technology at a low dose and low contrast agent flow rate/dosage with traditional CCTA images, while exploring the potential application of this technology in early screening of coronary artery disease.

Methods: From March 2024 to September 2024, 60 patients suspected of having coronary artery stenosis were enrolled in this study. Ultimately, 46 patients were included for analysis. Based on different acquisition protocols, divided into Group A and Group B. Group A underwent conventional computed tomography (CT) angiography with a tube voltage of 120 kV, a contrast agent injection rate of 6 mL/s, and a dosage of 0.9 mL/kg. Group B received a triple-low CCTA protocol with a tube voltage of 100 kV, a contrast agent injection rate of 2 mL/s, and a dosage of 0.3 mL/kg. Additionally, Group C was created by applying CE-Boost combined with a deep learning reconstruction technique to Group B images. The radiation dose and contrast agent dosage were compared between Group A and Group B. The image quality of the three groups, including CT values, background noise, signal-to-noise ratio (SNR), and contrast signal-to-noise ratio (CNR), was also compared, with p < 0.05 indicating significant statistical differences.

Results: Our results indicate that Group A required 67.8% more contrast agent and a 52.0% higher radiation dose than Group B (64.68 ± 3.30 mL vs. 20.19 ± 2.22 mL, 6.21 (4.60, 7.78) mSv vs. 2.05 (1.42, 4.33) mSv, all p < 0.05). Image analysis revealed superior subjective scores in Groups A (4.68 ± 0.72) and C (4.38 ± 0.95) versus Group B (4.25 ± 0.10) (both p < 0.05), with no statistical difference between Groups A and C. CT values were significantly elevated in Group A across all vessels compared to both Groups B and C (p < 0.05), while Group C exceeded Group B post CE-Boost. SNR comparisons showed Group A dominance over B in the proximal right coronary artery (RCA-1)/left main coronary artery (LM)/left anterior descending coronary artery (LAD)/left circumflex coronary artery (LCX) and over C in the RCA-1/LM (p < 0.05), contrasting with the superiority of SNR in Group C versus B in the middle right coronary artery/distal right coronary artery (RCA-2/3)/LM/LAD/LCX. CNR analysis demonstrated an equivalent performance between A and C, though both groups significantly surpassed Group B (A vs. B: p < 0.05; C vs. B: p < 0.05).

Conclusion: The triple-low CCTA protocol using CE-Boost technology combined with deep learning reconstruction, achieved a 52% reduction in radiation exposure and a 67.8% reduction in contrast agent usage, while preserving diagnostic image quality (with CNR and noise levels comparable to standard protocols). This demonstrates its clinical feasibility for repeated coronary evaluations without compromising diagnostic accuracy.