Robust Estimation of Kinetic Parameters in Dynamic PET Imaging

Dynamic PET imaging provides important information for biological research, clinical diagnosis and pharmacokinetic analysis through kinetic modeling and data-driven parameter estimation. Kinetic parameters quantitatively describe dynamic material exchange and metabolism of radiotracers in plasma and tissues. While many efforts have been devoted to estimate kinetic parameters from dynamic PET, the poor statistical properties of the measurement data in low count dynamic acquisition and the uncertainties in estimating the arterial input function have limited the accuracy and reliability of the kinetic parameter estimation. Additionally, the quantitative analysis of individual kinetic parameters is not yet implemented. In this paper, we present a robust kinetic parameter estimation framework which is robust to both the poor statistical properties of measurement data in dynamic PET and the uncertainties in estimated arterial input function, and is able to analyze every single kinetic parameter quantitatively. The strategy is optimized with robust H-infinity estimation under minimax criterion. Experiments are conducted on Monte Carlo sirnulated data set for quantitative analysis and validation, and on real patient scans for assessment of clinical potential.