Simulated FDG-PET studies for the assessment of SUV quantification methods

Abstract

Aim: To study in detail the accuracy and repeatability of three commonly used methods for SUV estimation in solitary pulmonary nodules.

Material and methods: We have designed a realistic framework based on simulated FDG-PET acquisitions from an anthropomorphic activity model that included solitary pulmonary nodules (different sizes) of well-known SUV. This framework enables us to compare the SUV values obtained from the reconstructed PET images with the real SUV values. Three commonly used methods (SUVmax, SUVmean and SUV50) were used to estimate the tumor activity.

Results: Our results showed the tumor activity was overestimated using SUVmax and clearly subestimated using SUVmean. Instead, the quantification of SUV50 showed great agreement with the simulated tumor activity and only slight subestimation was found for very small lesions. On the other hand, SUVmean showed better performance than SUV50 in terms of repeatability, providing variabilities below 5% for all tumor sizes and for injected doses as low as 111 MBq.

Conclusions: Our findings showed that SUV50 provided better performance for estimating accurately tumor SUV values in pulmonary nodules, but SUVmean showed better results in terms of repeatability.

Keywords: Cuantificación; Monte Carlo simulation; PET; Quantification; SUV; Simulación Monte Carlo.