Comparison of iterative and statistical cone-beam CT reconstruction methods

The goal of this master thesis has been to compare iterative and statistical algorithms in the context of reconstruction methods for Cone Beam Computed Tomographies (CBCT). These families of algorithms are more efficient with limited data and more robust against artifacts in comparison with the standard reconstruction technique: FDK. The paper starts with an introduction to the context and the use of CBCTs, with special attention to the applications within protontherapy. The next section presents the reconstruction algorithms that could be used with CBCTs: filter back projection algorithms, algebraic reconstruction techniques, TV-based algorithms and statistical algorithms. The third section deals with the method used to produce and asses the reconstructions. It also introduces some image quality metrics. A detailed explanation of the TIGRE toolbox is also part of that section. Finally the fourth and fifth sections are focusing on the results produced by the tested reconstruction algorithms and the discussion of these results. The discussion emphasises the comparison of the algorithms with FDK and discusses whether or not they could be a realistic alternative.