Improving the image quality of coronary CT angiography using a novel non-rigid registration based motion correction algorithm: in comparison with bi-sector reconstruction

Objective: Intelligent boundary registration (IBR) is a novel non-rigid registration based motion correction algorithm and is recently introduced for coronary CT angiography (CTA). The aim of this study was to compare the image quality of IBR with that of bi-sector reconstruction in coronary CTA. Methods: Coronary CTA was performed on 70 patients with mean heart rates of 65 bpm (range: 58-75 bpm) using a 64-row MDCT scanner. Bi-sector multiphase reconstructions were performed to select the optimal cardiac phase. Single sector reconstruction was used to select the optimal cardiac phase. IBR reconstruction was then performed based on the optimal single sector reconstruction to generate the IBR images. The image quality was evaluated by two blinded and independent observers on the workstation using a 5-point scale (5 = excellent to 1 = non-diagnostic). The image quality of the two reconstruction methods was compared. Evaluation criteria included overall image quality, continuity, motion artifacts, blurred or sharp edges of blood vessels. Results: A total of 984 coronary artery segments from 70 patients was analyzed. There was good inter-observer agreement (k > 0.81). The average score of IBR (4.45 ± 1.01) was significantly higher (Z = -9.22, P < 0.01) than that of bi-sector reconstruction (4.02 ± 1.28). On coronary artery segment analysis, the percentage of acceptable image quality score ≥ 3 or ≥ 4 was significantly higher (χ² = 5.68, P < 0.05; χ² = 6.47, P < 0.05) with IBR (91.2%, 88.5%) than bi-sector reconstruction (88.3%, 73.3%). IBR reconstruction showed significantly lower (χ² = 12.13, P < 0.01) non-assessable segments (4.2%) than bisector reconstruction (9.5%). Conclusion: IBR improves the image quality of coronary CTA, especially in patients with stair-step artifacts. (authors)