Positron Emission Tomography (PET) is an increasingly used imaging procedure, mostly used in clinical oncology, neurology and other medical diagnostics areas as well as in pre-clinical research, particularly in cancer research. The process of the PET imaging is complex, needs high computing capacity, but nowadays it is getting cheaper, data evaluation time is getting shorter, and results are more and more accurate, hence PET imaging equipment is getting widespread.
Reconstructing the image from the radiation data collected in PET examination is much harder than in traditional imaging processes such as CT and MRI, because we need to work on much less quantity of much noisier data, which needs detailed investigation and simulation of the phenomena occurring in the process.
In my thesis I present the PET reconstruction process, especially one iteration type of the reconstruction algorithm, the forward projection, and the most necessary theories for understanding the algorithm. I implement the forward projection process on graphical hardware and analyse the results obtained from different forward projection operators.