Impact of different PET reconstruction algorithms on18F FDG SUV and MTV of tumor lesions

Objectives: To analyze the impact of Bayesian penalized likelihood (BPL) PET/CT reconstruction algorithm on 18F-FDG quantitative parameters of malignant tumor lesions.

Methods: The 18F-FDG PET/CT images of 76 patients with malignant tumors were retrospectively analyzed, using BPL (TOF+PSF+BPL) and non-BPL (TOF+PSF) algorithm for reconstruction. The 18F-FDG quantitative parameters were measured respectively, including: SUVmax, SUVmean, SBR (SBR= lesion SUVmax/liver SUVmean), MTV of tumor lesions, and calculated the change value of quantitative parameters (such as Change_SUV=SUV_BPL-SUV_non-BPL).

Results: A total of 196 malignant tumor lesions, SUVmax 11.06 (8.01, 14.48) g/ml, SUVmean 6.90 (5.23, 9.15) g/ml, SBR 5.03 (3.86, 6.76) in BPL group were significantly higher than non-BPL group 9.14 (6.36,12.51), 5.44 (3.79, 7.63) and 4.29 (3.08, 5.63) (Z=4.04, 4.65 and 3.98, all P<0.0011); but the MTV 1110 (480.00, 2468.50)ml in BPL group was significantly lower than 1345 (734.00, 2710.50)ml of non-BPL group (Z=-2.45, P<0.05). Change_SUVmax, Change_SUVmean, Change_SBR and Change_MTV in <10mm group were significantly greater than those in ≥10mm group (Z=6.745,7.739,6.454 and -0.757,all P<0.05); Change_SUVmean in SUVmax≥10 group was significantly higher than SUVmax<10 group (Z=-2.042, P<0.05), while Change_MTV was the opposite. Change_MTV of SUVmax<10 group was significantly greater than that of SUVmax≥10 group (Z=-3.162,P<0.05).

Conclusions: This study clarifies that BPL can increase tumor SUV, especially <10mm and SUVmax≥10 lesions. The SUV threshold for the differential diagnosis of benign and malignant under BPL reconstruction needs to be adjusted appropriately. At the same time, it is proposed that BPL significantly reduces tumor MTV, with <10mm and SUVmax<10 lesions as obvious, and the interpretation criteria of 18F-FDG PET/CT quantitative parameter for treatment evaluation under BPL reconstruction needs further study to avoid overestimation of tumor burden. Keywords: Tumor; Positron-emission tomography; Tomography; X-ray computed; Image processing; Deoxyglucose; Quantification