Motion compensated iterative reconstruction for cardiac X-ray tomography

textabstractWithin this Ph.D. project, three-dimensional reconstruction methods for moving objects (with a focus on the human heart) from cone-beam X-ray projections using iterative reconstruction algorithms were developed and evaluated. This project was carried in collaboration with the Digital Imaging Group of Philips Research Europe – Hamburg. In cardiac cone-beam computed tomography (CT) a large effort is continuously dedicated to increase scanning speed in order to minimize patient or organ motion during acquisition. In particular, motion causes severe artifacts such as blurring and streaks in tomographic images. While for a large class of applications the current scanning speed is sufficient, in cardiac CT image reconstruction improvements are still required. Whereas it is currently feasible to achieve stable image quality in the resting phases of the cardiac cycle, in the phase of fast motion data acquisition is too slow. A variety of algorithms to reduce or compensate for motion artifacts have been proposed in literature. Most of the correction methods address the calculation of consistent projection data belonging to the same motion state (gated CT reconstruction). Even if gated CT leads to better results, not only with respect to the processing time but also regarding the image quality, it is also limited in its temporal and spatial resolution due to the mechanical movement of the gantry. This can lead to motion blurring, especially in the phases of fast cardiac motion during the RR interval. A motion-compensated reconstruction method for CT can be used to improve the resolution of the reconstructed image and to suppress motion blurring. Iterative techniques are a promising approach to solve this problem, since no direct inversion methods are known for arbitrarily moving objects. In this work, we therefore introduced motion compensation into image reconstruction. In order to determine the unknown cardiac motion, 3 different cardiac-motion estimation methodologies were implemented. Visual and quantitative assessment of the method in a number of applications, including: phantoms; cardiac CT reconstructions; Region of Interest (ROI) CT reconstructions of left and right coronaries of several clinical patients, confirmed its potential