Improved image quality with simultaneously reduced radiation exposure: Knowledge-based iterative model reconstruction algorithms for coronary CT angiography in a clinical setting.

Background: The aim of this study was to assess the potential for radiation dose reduction using knowledge-based iterative model reconstruction (K-IMR) algorithms in combination with ultra-low dose body mass index (BMI)-adapted protocols in coronary CT angiography (coronary CTA).

Methods: Forty patients undergoing clinically indicated coronary CTA were randomly assigned to two groups with BMI-adapted (I: <25.0 kg/m, II: <28.0 kg/m, III: <30.0 kg/m, IV: ≥30.0 kg/m) low dose (LD, I: 100kV/75 mAs, II: 100kV/100 mAs, III: 100kV/150 mAs, IV: 120kV/150 mAs, n = 20) or ultra-low dose (ULD, I: 100kV/50 mAs, II: 100kV/75 mAs, III: 100kV/100 mAs, IV: 120kV/100 mAs, n = 20) protocols. Prospectively-triggered coronary CTA was performed using a 256-MDCT with the lowest reasonable scan length. Images were generated with filtered back projection (FBP), a noise-reducing hybrid iterative algorithm (iD, levels 2/5) and K-IMR using cardiac routine (CR) and cardiac sharp settings, levels 1-3.

Results: Groups were comparable regarding anthropometric parameters, heart rate, and scan length. The use of ULD protocols resulted in a significant reduction of radiation exposure (0.7 (0.6-0.9) mSv vs. 1.1 (0.9-1.7) mSv; p < 0.02). Image quality was significantly better in the ULD group using K-IMR CR 1 compared to FBP, iD 2 and iD 5 in the LD group, resulting in fewer non-diagnostic coronary segments (2.4% vs. 11.6%, 9.2% and 6.1%; p < 0.05).

Conclusions: The combination of K-IMR with BMI-adapted ULD protocols results in significant radiation dose savings while simultaneously improving image quality compared to LD protocols with FBP or hybrid iterative algorithms. Therefore, K-IMR allows for coronary CTA examinations with high diagnostic value and very low radiation exposure in clinical routine.