Abstract
3186
Introduction: The total body PET scanners such as uEXPLORER have been developed with higher temporal resolution and a more extended axial field of view. When employing traditional quantitative methods like the compartmental model for total-body imaging on such scanners, a single kinetic model may not be feasible for multiple organs and the importance of organ-dependent delay time has been shown in recent studies [G Wang et al 2021, T Feng et al 2021]. The objective of this study is to assess the feasibility of various compartmenal models and delay estimation schemes for diverse tissues.
Methods: One, irreversible and reversible two tissue compartmental models (1C, 2Ci, 2Cr) with three (K1, k2, Vb), four (K1, k2, k3, Vb) and five (K1, k2, k3, k4,Vb) parameters are all applied. The time delay (td) for different organs is jointly estimated with above parameters in each model. Two schemes are considered: (i) td is determined by early kinetics (first 5 min) using 1C model; (ii) td can also be recovered by full-time data (one hour) in arbitrary models. The performance of these models and delay schemes are evaluated in a lung adenocarcinoma patient study. It was conducted on the uEXPLORER with the approval of the Ethics committee of Henan Provincial People's Hospital. A 60-min dynamic scan was performed immediately after the administration of 18F-FDG and reconstructed as following sequences: 50x2s, 20x10s, 10x30s, 10x60s, 8x300s. A series of regions of interest (ROIs) in diverse organs/tissues (liver, spleen, kidney, lung, bladder, bone, tumor, myocardium and brain) were drawn manually. The input function is obtained from a ROI placed in the descending aorta. 1C, 2Ci and 2Cr are used to fit the time activity curves from these ROIs successively. Associated parameters and delay terms are derived. The simple 1C model is implemented again to the first 5-min data for estimating the optimal td efficiently.
Results: The time activity points and fitted curves with 1C, 2Ci and 2Cr for all main organs are presented (Fig. 1). Results show that 2Cr has the advantages for liver, spleen, lung, bone, tumor and brain and 2Ci fits best for the kidney. The goodness of fits for myocardium with three models are very similar. For bladder, the inapplicability of compartmental models (1C, 2Ci and 2Cr) is apparent. Delay times obtained from different models and schemes are also reported (Fig. 2). They are slightly different for liver, spleen, kidney and lung, but the estimation from 1C with a first 5 min data is very close to that from 2Ci or 2Cr for most of the organs, except bladder.
Conclusions: The feasibility of the compartmental models for the bladder is questionable. An efficient delay estimation scheme using 1C with early time frames data is validated in this study. A non-parametric residue mapping approach (Gu, O’Sullivan et al. 2021, O’Sullivan et al. 2009) will be considered to further enhance the kinetic quantitation as it has some important features such as the flexibility for diverse tissue environments, consideration of organ-dependent delays and also the ability to address issues with bladder.