SPARC: Sectorized Partial-Angle Reconstruction Based Motion Correction Method for Multi-Phase CCTA

Abstract Objective. Multi-phase Coronary Computed Tomography Angiography (CCTA) provides a clear visualization of coronary arteries, offering comprehensive and reliable imaging support for the clinical diagnosis of coronary artery disease. However, traditional motion correction reconstruction algorithms lack reconstruction robustness despite their interpretability, while deep learning reconstruction methods achieve high imaging quality at the expense of interpretability and anatomical reliability in clinical CCTA practice. This study aims to simultaneously achieve interpretable and robust multi-phase coronary reconstruction. Approach. We propose a motion trajectory estimation and motion correction framework termed sectorized partial-angle reconstruction-based motion correction (SPARC) for CCTA. The proposed framework integrates sectorized partial-angle reconstructions, a spatio-temporal cascaded network for coronary centerline extraction, and a phase-accumulated fuzzy C-means clustering algorithm for motion trajectory estimation. These components work together to minimize motion estimation errors and enhance the quality of motion-corrected reconstruction. Main results. Experimental results from both simulations and clinical studies demonstrate that SPARC achieves superior reconstruction performance compared with existing methods. SPARC effectively reduces coronary motion artifacts and restores clear vessel structures while preserving reconstruction fidelity, achieving the best scores on all metrics. Significance. By preserving both interpretability and anatomical reliability while effectively reducing motion artifacts, SPARC provides a clinically deployable solution for motion-corrected multi-phase CCTA, supporting more reliable imaging for coronary artery disease diagnosis.