Framework to quantify the metabolic rate in the heart using Monte Carlo simulation and compartmental modeling

摘要

The Nuclear Medicine imaging using PET modality allows evaluating the physiological condition of the heart (i.e. heart disease, perfusion, cardiac stress, physics disorders, etc.), by the use of small amounts of radioactive material. While the visual information provided by PET image is quite important and today is heavily used by the specialists, it is clear that having quantitative values of the metabolism would help to determine the real condition of the myocardium, to get a more accurate diagnosis and to apply a better treatment. Thus, we developed a framework using different tools to simulate a real PET exam focused in the heart in order to analyze the metabolic exchange in this organ. This research produced realistic PET exams by using GATE platform that performs Monte Carlo simulations, together with an anthropomorphic phantom of the whole body called MASH. This study consisted of the modeling of a commercial PET scanner with BGO detectors, where the MASH's thorax was imaged by the scanner. In order to enhance the realism of the simulation, clinical data was considered, like the quantity of radiotracer (FDG) injected, time of acquisition, and number of frames along the time, among others. Next, the projections were reconstructed using STIR's algorithms. With the aim of obtaining a better quality of the volumes and to select the specific ROI, filtering and segmentation algorithms were applied, respectively. Finally, the heart metabolic analysis was performed using a mathematical model that seeks to describe and to quantify the level of consumption and exchange of glucose in the heart through a model with three compartments. The metabolic parameters obtained were: K1=0.5690, k2=0.2266, k3=0.0718, and k4=0.0243. It was evidenced that the process of metabolic quantification using compartmental modeling is significantly relevant because of its flexibility, noninvasiveness and reliability. Hereafter, we will apply this framework under real PET images.