Respiratory displacement modelling in cone beam computed tomography

Abstract

Any organ motion during the projection acquisition causes blurring artefacts in the Computed Tomography (CT) reconstructions. It is a drawback in applications of IntroductionOne of the problems encountered in the dynamic volumetric computed tomography (CT) based imaging is the motion due to respiration. If neglected, it causes artifacts in the tomographic reconstruction.Isocentric imaging equipment, such as in-theatre Carm units and radiotherapy simulators, equipped with high quality flat panel detectors, are potential platforms for cone beam CT (CBCT) imaging applications. However, due to limitations concerning the C-arm rotating speed and the maximum detector readout rate, the complete scanning cannot be accomplished within a single breath hold and application of respiratory motion compensation is required.Several algorithms for respiratory motion correction have been proposed in the literature. Crawford et al. [1,2] proposed a model-based method where the respiration, in an object cross-section, is modelled by time varying magnification and displacement. Ritchie et al. [3] refined their method, assuming that the motion model is correct only in a small region around each object, and developed a pixel-specific filtered backprojection algorithm. Wang et al.[4] proposed a knowledge-based helical multi-slice/CBCT approach for dynamic volumetric imaging. Their approach uses the time varying anatomical information of the heart and is suitable for CBCT. McLeish et al.[5] derived a patientspecific model of the respiratory movement and the deformation of the heart's surface for the MRI imaging. It is based on relatively low-resolution pre-scans with breath holds in different positions of the respiratory cycle, and monitoring landmarks on the reconstructed surface of the heart.Our study concerns CBCT with scanning over circular trajectory and concurrent respiratory monitoring. The objectives of this study are to develop a model for the displacement of the organ of interest during the respiratory cycle, and to apply it in the tomographic reconstruction algorithm, in order to account for organ movement due to respiration.Considering heart imaging, using simulations, we were able to demonstrate that its respiratory displacement can be estimated from the same projection data, required for a CBCT reconstruction. Our work includes semi-automatic segmentation of the heart on the projection images, reconstruction of a 3D-heart object that performs the same motion as the heart during respiration and applying this information in the CBCT reconstruction algorithm. The results reveal significantly improved quality of the computed cardiac tomograms.Although the feasibility of this approach has been demonstrated for heart imaging, it is generic and applies to imaging of any other organ, with a convex hull, that is subject to respiratory displacement during scanning. MethodsThe work includes simulation of the acquisition of projection images, deriving a model of the heart motion due to respiration, modifying a CBCT reconstr...