Vorschau

PDF, Englisch Download (14MB) | Nutzungsbedingungen

Abstract

Motion artifacts in cardiac CT remain an issue that can degrade image quality. In addition to artifacts originating from cardiac motion in general, irregular motion in particular may lead to unique artifacts for scanners with partial coverage of the heart. If multiple sub-volumes, or stacks, have to be reconstructed to cover the heart, irregular motion may introduce discrepancies between these stacks. The resulting artifacts in the finally assembled CT volume are, herein, referred to as stack transition artifacts. A stack transition artifact removal (STAR) method based on symmetric image registration was developed. The registration computed a smooth motion vector field using data from the overlap of neighboring stacks. The extent of the vector field smoothing was set automatically for individual registration tasks. Furthermore, a motion compensation (MoCo) was developed to address the general motion artifacts in short-scan data. This method is based on partial angle reconstructions to which a 4D motion model is applied. The latter was optimized using a cost function consisting of a penalized image entropy. The penalties refer to cardiac velocity and strain. Both methods were applied to single-phase reconstructions from clinical data. In addition, simulations were used with STAR. STAR considerably improved image quality. Discontinuities were removed or clearly reduced. The MoCo improved image quality for all artifact impaired reconstructions. In case of artifactfree images, image quality was maintained, however, a tendency for a minor blurring in the motion-compensated images was observed. Finally, both artifact correction methods showed a capability to clearly improve image quality.