Massively Parallelizable Reconstruction for High Energy Physics and Medical Imaging

My PhD research activity was funded by the Istituto Nazionale di Fisica Nucleare (INFN) and by the Physics Department of the University of Pisa. The goal of the research activity is to face the problem of the reconstruction by exploiting two different kinds of parallelism in two different research fields: high energy physics (HEP) and medical physics. Reconstruction can be a very challenging computational task in both fields. As regards HEP experiments, I worked to the optimization of the Fast Tracker (FTK) processor for the ATLAS experiment at the Large Hadron Collider (LHC) located at the CERN laboratory in Geneva. ATLAS is one of the two general purpose detectors designed to measure the product of the proton-proton collisions at the LHC. The key role in the novel technology is played by the Associative Memory (AM) that is a massively parallel system used to provide very fast online event selection of the images of the particles emerging from the collisions of protons. As regards medical physics, I developed a parallel implementation of an iterative algorithm for image reconstruction in 3D Positron Emission Tomography (PET) using Graphics Processing Units (GPUs). The implementation of the iterative algorithms can be a very challenging task due to the massive amount of computation required to incorporate accurate system modeling. For this reason, GPUs are currently used to attain reconstructed images in a practical time because they high performance supporting massive parallel computing power