Assessing Image Quality and Diagnostic Performance of Quadruple-Low Coronary CT Angiography with Deep Learning in High-BMI Patients.
Authors
Affiliations (3)
Affiliations (3)
- The Department of Radiology, The First Affiliated Hospital of Zhengzhou University, No.1, Eastern Jianshe Road, Zhengzhou, Henan 450052, China (K.Q., S.L., J.Z., X.W., P.H., D.Y., Y.Z., A.D., Q.T., M.Z., W.Z., J.G., J.L.).
- CT Imaging Research Center, GE HealthCare China, Beijing, China (L.W.).
- The Department of Radiology, The First Affiliated Hospital of Zhengzhou University, No.1, Eastern Jianshe Road, Zhengzhou, Henan 450052, China (K.Q., S.L., J.Z., X.W., P.H., D.Y., Y.Z., A.D., Q.T., M.Z., W.Z., J.G., J.L.). Electronic address: [email protected].
Abstract
To examine the feasibility of a quadruple-low protocol in coronary computed tomography angiography (CCTA) assisted by the deep learning image reconstruction (DLIR) for participants with high body mass index (BMI). This prospective study involved 180 participants (BMI of ≥ 25 kg/m<sup>2</sup>). Participants were randomly assigned to three groups: the standard-dose (SD) group in 100 kVp with adaptive statistical iterative reconstruction (ASIR-V 50%) using contrast media (CM) administration at an iodine load of 245 mgI/kg (Iohexol, 350 mgI/mL) with a 10 s injection time; the experimental group underwent imaging with 80 kVp and DLIR-H, using an iodine load of 128 mgI/kg with a 7 s injection time, representing the triple-low (TL) with 350 mgI/mL Iohexol and quadruple-low (QL) groups with 320 mgI/mL Iodixanol. Quantitative image quality evaluations included vascular attenuation, image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR), while qualitative image quality and diagnostic performance were also assessed. Compared to the SD group, the TL group achieved reductions of 29.4% in radiation dose, 47.5% in CM volume, and 23.5% in CM injection rate, while the QL group achieved reductions of 32.5% in radiation dose, 41.5% in CM volume, 14.5% in CM injection rate, and 9% in CM concentration. The image noise in the QL and TL groups were significantly lower than in the SD group (all p < 0.001). Overall, the SNR, CNR, and qualitative image quality of the TL and QL groups were superior to those of the SD group (p < 0.05). The QL group maintained the stenosis diagnostic performance (all p > 0.05). TL and QL CCTA protocols showed equivalent or superior image quality and diagnostic accuracy compared to the SD protocol and may serve as effective alternatives for reducing radiation dose and contrast media use in patients with high BMI.