Parametric Maps of Kinetic Heterogeneity and Ki in Dynamic Total Body PET using Approximate Bayesian Computation

The quantification of [18F]Fluorodeoxyglucose (FDG) metabolism in oncology via positron emission tomography (PET) is essential for accurate diagnosis and treatment planning. Traditional semi-quantitative methods like the Standard Uptake Value (SUV) and Patlak plots, although widely used, often fall short in addressing the complexities of tissue heterogeneity. We introduce a voxelwise Approximate Bayesian Computation (ABC) for total body PET (TB-PET) imaging heterogeneity study and kinetic parameter estimation, demonstrating high accuracy in simulation and distinguishing reversible from irreversible FDG phosphorylation using real TB-PET data. The study successfully computed model probability maps and kinetic parameters (Ki) across an extensive real dataset of over 44 million voxels, achieving a significant computational speed-up. Future work will expand on these findings by exploring additional heterogeneities and their clinical implications, as well as incorporating spatial correlations.