Statistical iterative reconstruction and dose reduction in multi-slice computed tomography

Computed tomography is one of the most important imaging methods in medical technology. Although computed tomography examinations only make up a small proportion of X-ray examinations, they do make a great contribution to civilizing radiation exposure of the population. By using statistical iterative reconstruction methods, it is possible to reduce the mean radiation dose per examination. While statistical iterative reconstruction methods enable the modeling of physical imaging properties, the user can decide freely and independently about the choice of numerous free parameters. However, every parameterization decision has an influence on the final image quality. In this work, inter alia the definition of the modeling of the forward projection is examined as well as the influence of statistical weights and data redundancies in interaction with various iterative reconstruction techniques. Several extensive studies were put together, which challenge these different combinations in every respect and push the models to their limits. Image quality was assessed using the following quantitative metrics: basic metrics and task-based metrics. The investigation shows that the definition of iterative reconstruction parameters is not always trivial and must always be understood comprehensively to obtain an optimal image quality. Finally, a novel reconstruction algorithm, called FINESSE, is presented, which improves some of the weaknesses of other reconstruction techniques.