Automatic cardiac phase point selection for 3D rotational coronary angiography

A fully automated projection-based method to determine one or more cardiac phases of least global motion for rotational coronary angiography is presented. This method is able to support gated 3D reconstruction of coronary arteries in the optimisation of the phase point, since the image quality resulting from gated reconstructions is optimal in cardiac phases of low motion. The algorithm is based on direct analysis of the acquired rotational X-ray projections. The motion function is extracted by cross-correlation of line integral profiles from adjacent projections. Based on the extracted motion function, the cardiac phase of least motion is determined for the gating window shape and width chosen in the 4D reconstruction. Gated circular filtered back-projection is used for 3D reconstruction. The feasibility of the approach is evaluated on 27 clinical cases, comprising both LCA and RCA acquisitions. All cases have been reconstructed at all distinguishable cardiac phases using a gating window width of 20% RR and cosine squared weighting. The reconstruction quality at the cardiac phases proposed by the presented algorithm is compared to the best phases of the full 4D sequence. It is shown that in all cases, at least one of the extracted phase candidates yields optimal reconstruction quality with respect to the optimal cardiac phases chosen by visual inspection.