Método analítico de reconstrução de imagens de SPECT em Python: aplicação da retroprojeção filtrada (FBP) em phantom de Jaszczak

Abstract

SPECT is a clinical imaging modality with proven value for diagnosing disease and monitoring treatments. Using specialized mathematical algorithms, the data obtained by SPECT make it possible to reconstruct the spatial distribution of radiopharmaceuticals in the region of interest, offering physiological information relevant to medical diagnosis. The purpose of this work was to implement in Python the analytical reconstruction method used in SPECT equipment, known as filtered back projection (FBP). The project includes the reconstruction of the physical Jaszczak phantom and the digital Shepp-Logan phantom. In this process, the acquisition geometry of the phantom projections and their distinctions were analyzed, resulting in the sinogram. Furthermore, all the necessary steps were implemented to reconstruct the object from these projections, including the use of the Hamming filter in the frequency domain. The implementation of the algorithm made it possible to reconstruct the slices of both phantoms, Jaszczak and Shepp-Logan, with reasonable success. During this process, all the steps that allow FBP to reconstruct the object from projections were analyzed, including problems arising from the technique, such as the appearance of artifacts and static noise. The FBP Python code was made available in full, allowing future studies on its limitations, improvements and artifacts in the technique.