Impact of point spread function reconstruction on quantitative 18F-FDG-PET/CT imaging parameters and inter-reader reproducibility in solid tumors.

INTRODUCTION: This study aims to determine the impact of point-spread function (PSF) reconstruction on quantitative PET/computed tomography (CT) indices and the inter-reader reproducibility of these measurements. MATERIALS AND METHODS: The study was approved by the Institutional Review Board under a waiver of informed consent. A total of 42 oncology patients with 85 lesions (all ≥ 2 cm) were included. The PET/CT images were reconstructed with PSF (OSEM+TOF, 2i, 21s, all-pass filter) and without PSF (OSEM+TOF, 2i, 21s, 5 mm Gaussian). For each lesion, the maximum, mean, and peak standardized uptake values (SUV), total lesion glycolysis (TLG), and metabolic tumor volume (MTV) were measured by two readers (R1 and R2) using a semiautomatic gradient segmentation method. Intraclass correlation coefficient (ICC) and Bland-Altman analyses were performed. RESULTS: There was excellent correlation between non-PSF and PSF reconstruction PET/CT values (ICC ≥ 0.96 for all parameters, P < 0.0001). Comparison of PSF with non-PSF images showed a mean bias (percentage change) of +11.97% (R1) and +11.94% (R2) for SUV max, +7.63% (R1) and +7.82% (R2) for SUV mean, +7.45% (R1) and +7.37% (R2) for SUV peak, -0.82% (R1) and -0.1% (R2) for TLG, and -6.68% (R1) and -5.65% (R2) for MTV. PSF reconstruction resulted in a decrease in MTV in 77.6% (R1) and 83.5% (R2) of lesions. Percentage changes in PSF versus non-PSF indices were not related to the site of the lesions (P > 0.05). Close agreement was observed between two readers (ICC ranged between 0.9 and 1.0, P < 0.0001). CONCLUSION: The PSF reconstruction increased the SUV max, SUV mean, and SUV peak, as expected, whereas it tended to produce lower values for MTV and had variable effect on TLG. This can be attributed to the ability of PSF reconstruction to better discern tumor uptake from activity spill-out.

Duke Scholars

Author Rathan Markandan Subramaniam Radiology