Advisors

Medical Image Reconstruction

Medical Image Reconstruction produces volumetric (3-D) information supporting physicians in diagnoses, planning, and treatment. For example, a Computed Tomography (CT) scanner acquires many x-ray images around the patient. The 3-D information is then reconstructed from these projections. In general, the term reconstruction describes the inversion of a transformation and also applies to other imaging modalities. In CT, the 3-D image is computed by inverting the x-ray projection scan.

Research at the LME:

Motion Compensation. Cardiac C-arm CT is used to generate 3-D images of the heart during interventions. This information is very valuable, e.g. for pacemaker implantations. A reconstruction of cardiac chambers can support the estimation of physiological parameters like ejection volume or heart wall motion. We estimate cardiac motion to do a dynamic reconstruction using motion correction. Benefits are increased temporal and spatial resolution due to reduced artifacts. Beam hardening artifacts severely degrade image quality and diagnostic accuracy, esp. for multi-material objects. The polychromatic x-ray spectrum causes nonlinear characteristics which make correction difficult. This research project aims to develop a correction algorithm which can effectively reduce beam hardening artifacts with less prior knowledge and higher computational efficiency. Perfusion C-arm CT is a novel technology to measure the capillary bloodflow. It can be used for diagnosis of stroke and help physicians to assess the success of interventional stroke procedures. Our research aims to provide a reliable reconstruction of CBF from slowly rotating C-arm systems. Molecular Imaging visualizes molecular processes in vivo for diagnosis and therapy. A tomographic 3-D image of the distribution of a radiotracer in the patient body is obtained. Quantitative SPECT seeks to estimate the activity distribution in the body in absolute terms. This requires knowledge of and correction for all the degrading factors in the imaging chain. Compressed Sensing. 4-D CT imaging supports treatment simulation and planning in radiotherapy. In presence of motion, only a limited number of projections can be used for reconstruction. With traditional methods, image quality is degraded. We work on novel reconstruction methods based on compressed sensing theory to reduce artifacts and improve the reconstruction accuracy.

Left: Perfusion map of digital brain phantom. Right: Volumetric image of radiotracer distribution in target organs, overlaid with a CT image.

Contact:

joachim.hornegger@informatik.uni-erlangen.de