Multicenter comparison of dose calibrator accuracy for PET imaging using a standardized source

Objectives The quantification accuracy of PET images largely depends on the accuracy of dose calibrators (DC) in measuring the radioactivity of either F-18 water or Ge-68 epoxy filled phantoms, which are not traceable to a NIST standard. This study evaluates the DC accuracy in measuring the radioactivity of a new Ge-68/F-18 NIST direct traceable standard to determine the calibration error of PET scanners.

Methods Three centers and 39 DCs (12 models) were included in this study. For each DC, the NIST standard was measured using the manufacturer's setting for Ge-68 and F-18 . These values were then compared to the expected value of each NIST direct traceable source and the percent difference (%diff) was calculated. The mean±STD of the %diff among all DCs, models of DCs, and different centers were calculated.

Results For F-18 (Ge-68) setting, the mean±STD of the %diff across all DC was 6.93±3.59% (6.26±3.48%). The highest and lowest average %diff were 10.15% (9.78%) and 6.24% (5.48%) for the CRC-7 (CRC-15PET) and Victoreen (CRC-15W) DCs. The Atomlab100 DCs consistently underestimated the activity of the NIST standard while all other DCs overestimated it. For all DC, the %diff for F-18 was higher than Ge-68 except for CRC15PET. The variation in %diff among the three centers was within 2% for both Ge-68 and F-18.

Conclusions Current DC setting for measuring F-18 overestimates the true activity thereby affecting the accuracy of clinical PET images. The use of a direct traceable NIST standard can overcome this error by adjusting the F-18 calibration setting on these DC.