[Calculation Accuracy of Gross Tumor Volume at the Diaphragm Boundary Evaluated Using Respiratory-gated PET/CT].

Objective: The present study aimed to clarify gross tumor volume (GTV) contouring accuracy at the diaphragm boundary using respiratory-gated PET/CT.

Methods: The lung/diaphragm boundary was simulated using a phantom containing F solution (10.6 kBq/mL). Tumors were simulated using spheres (diameter, 11-38 mm) containing F and located at the positions of the lungs and liver. The tumor background ratios (TBR) were 2, 4, and 8. The phantom was moved from the superior to inferior direction with a 20-mm motion displacement at 3.6 s intervals. The recovery coefficient (RC), volume RC (VRC), and standardized uptake value (SUV) threshold were calculated using stationary, non-gated (3D), and gated (4D) PET/CT.

Results: In lung cancer simulation, RC and VRC in 3D PET images were, respectively, underestimated and overestimated in smaller tumors, whereas both improved in 4D PET images regardless of tumor size and TBR. The optimal SUV threshold was about 30% in 4D PET images. In liver cancer simulation, RC and VRC were, respectively, underestimated and overestimated in smaller tumors, and when the TBR was lower, but both improved in 4D PET images when tumors were >17 mm and the TBR was >4. The optimal SUV threshold tended to depend on the TBR.

Conclusions: The contouring accuracy of GTV was improved by considering TBR and using an optimal SUV threshold acquired from 4D PET images.