Objectives: This study investigates how sampling 18 FFDG (Fluorodeoxyglucose) at different time points (TPs) impacts personalised internal dosimetry in PET/CT imaging, with the goal of optimising multitracer PET protocols. The research focuses on evaluating different time-activity curve (TAC) fitting models to address gaps in current dosimetry practices. Material and Methods: Twenty patients underwent whole-body PET/CT scans after administration of 370 MBq 18 FFDG at four TPs: 5, 20, 60, and 90 min post-injection (p.i.). The liver, lungs, and spleen were selected as organs of interest. TACs for each organ were analysed using empirical and analytical (mono- and bi-exponential) models. Organ S-values were calculated using the GATE Monte Carlo (MC) simulator, and mean organ doses were estimated following the MIRD formalism. Results: The MC-calculated average S-value for the liver was 6.45E-05 mGy/MBq.s. Empirical TAC modelling overestimated internal doses, with significant bias observed for the spleen, where the dose overestimation reached 88% for sparse sampling. Bi-exponential TAC fitting showed superior accuracy, with 13% and 21% overestimation for the liver and spleen, respectively, compared to mono-exponential modelling. The lungs and liver showed comparatively lower bias in dose estimation. Conclusion: Bi-exponential TAC modelling provides greater accuracy than both mono-exponential and empirical models. Denser sampling schedules reduce bias across all models. Furthermore, acquiring the baseline PET/CT measurement at 60 min p.i. is recommended for precise individualised 18 F FDG dosimetry.
Karimipourfard et al. (Fri,) studied this question.