Abstract:Objective To investigate the performance of a knowledge-based iterative reconstruction (iterative model reconstruction - IMR) compared to filtered back projection (FBP) and a hybrid iterative reconstruction technique (iDose4) in coronary CT angiography under different tube voltages. Methods Five miniature pigs underwent prospective axial ECG-gated coronary 256-slice spiral CT angiography at different tube voltages including 120, 100 and 80 kV (group A, B and C). All original data were reconstructed using the FBP, iDose4 and IMR algorithms. The objective indices of the reconstructed images, including image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and mean CT value of the aortic root were compared among the three groups reconstructed by the different algorithms. A 5-point scale was used to assess subjective image quality of coronary segments and images with a score of ≥ 4 as a good criterion. Results At all three tube voltages, the IMR algorithm consistently yielded a lower noise and a higher CNR than iDose4 and FBP (P < 0.05). The differences of subjective scores among different reconstruction methods in the same group were all statistically significant (P < 0.05). In addition, at 80 kv, the image noise in IMR reconstruction was lower than in iDose4 and FBP reconstructions at 120 kV (P < 0.05); the radiation dose in group C was only 33.68% of that in group A. Conclusions Compared to iDose4 and FBP algorithms, iterative model reconstruction (IMR) algorithm significantly improves the image quality of 256-slice coronary CT angiography. A combination of low tube voltage (80 kV) and IMR algorithm can help reduce radiation dose and improve image quality.