Abstract: For coronary computed tomography (CT) angiography, the reconstruction phase finally used has to be iteratively adapted to the patient-specific heart motion to provide optimal image quality and therewith to improve the diagnostic value. The purpose was to provide an automatically raw data-based identification of the patient-specific optimal reconstruction phase for cardiac computed tomography imaging. We validated our method by a visual assessment of 65 patient data sets. In 52% of all cases, the highest correlation of the computed and the visually identified optimal reconstruction phase was ensured. In 30% of the cases, our method provided a higher image quality compared with the results obtained in routine clinical work. Our identification of the optimal reconstruction phase is a reliable method and can improve the medical workflow by providing optimal image quality with the initial image reconstruction, making multiple time-consuming image reconstructions obsolete.