Reconstruction of an image of a moving object from volumetric data

In the CT imaging of non-homogeneously moving objects such as the heart or the coronary vessel tree, there is a problem that different parts of the objects are at rest at different points in time. Thus, a gated reconstruction with a globally selected time point does not yield a sharp image of such objects. According to the present invention, a motion of the object is estimated, describing the motion of selected regions of these objects. Then, on the basis of the estimated motion, time points are determined, where these areas have minimal motion. Then, an image is reconstructed, wherein the data from which the respective regions are reconstructed, correspond to the respective time points, where the regions have minimal motion. Due to this, an improved image quality may be provided.