Hierarchical Dual-Tracer Modeling for Rapid [18F]FLT and [18F]FDG PET Scans on Mice with Lymphoma Tumors

374 Objectives The combination of FLT and FDG PET has shown clinical values in cancer diagnosis [1,2]. For rapid dual-tracer PET imaging, the fitting of highly complex dual tracer models suffers from local minima. This study employs hierarchical modeling [3] for rapid dual-tracer PET and tests the algorithm preclinically. Methods 4 mice with SUDHL-1 xenograft tumor were included. For each mouse, dynamic FDG PET was first acquired and a rapid dual-tracer PET (FDG at 0 min, FLT at 15 min) was scanned 2 days later. Each mouse was immobilized during the scans using an individualized board. 2 venous blood samples were taken after scanning and then measured using gamma counter and thin layer chromatography (TLC). A combined three exponential model was applied to obtain individual AIFs. Hierarchical modeling was applied by spatially subdivided into a number of clusters hierarchically. The initial values and fitting boundaries were refined hierarchically. Then pixel-wise pharmacokinetic modeling was performed with refined initial values and fitting boundaries. The correlation between model separated FDG images and the reference FDG images were investigated. The RMSE of hierarchical modeling was compared with the conventional pixel-wise fitting. Results The separated AIFs agreed with TLC. 1 mouse was failed for whole body registration and regional registration was replaced. The separated FDG images from dual-tracer scan showed good visual similarity to the reference FDG scans except for the heart, kidney and bladder regions. The SUV scatter plots of the separated and reference FDG images for the co-registered mice have significant correlations (r>0.6, p Conclusions Hierarchical modeling can improve the separation of rapid dual-tracer PET. This preclinical study demonstrated the feasibility for further clinical tests. Research Support DFG SFB 824