Abstract
1739
Objectives Track quantitative stability of a PET/CT scanner over a multi-year period, comparing counts, image activity, and other system parameters.
Methods We tracked measured counts and activity concentration (ρ) from images measured on a Gemini TF PET/CT scanner between 2013-2015 via daily, weekly, and monthly scans of a Na-22 point source, a 4.5 cm Ge-68 cylinder, and a 20 cm diameter F-18 water phantom. Stability analyses were performed on normalized count data corrected for isotope decay. We used SUV= (measured ρ / true ρ), to assess ρ. We reported SUV range and the COV=standard deviation/mean, across time. We also tracked: energy and timing resolutions, detector gains, temperatures inside the gantry, on electronics boards, and in the imaging suite.
Results Measured counts showed excellent stability, and ρ was piecewise stable over time depending on parameters in a reconstruction correction file. Coincidence counts measured daily from the Na-22 point source scans (Nscan=365, Nevents=7.8M-12M) and weekly from the Ge-68 phantom (Ns=81, Ne=2.5M-12M) had relative range (COV) of 92.8%-102.0% (1.05%) and 97.6%-102.3% (0.52%), respectively. A jump of ~6% in ρ was observed after an automated quantitative adjustment was incorrectly applied by a particular calibration protocol. Once the proper quantitative corrections were made, the relative range (COV) of ρ from the Ge-68 scans (Ns=49) was 98.4%-102.5% (0.97%). SUV in F18 water phantoms (Ns=17) ranged from 98.4% to 109% with COV of 3.47%. Other system and environment parameters were very stable making it difficult to assess their influence on scanner quantification.
Conclusions Measured counts and ρ from a fixed Ge-68 phantom were within ± 2.5% of the mean, with COV < 1%, using standard scanning methods. For refillable F-18 phantoms the relative range was ±5%. SUV mean and COV from a large volume of interest should be tracked with two significant figures to capture this level of quantitative consistency on PET scanners. Correlating variations in phantom SUV with measured counts could prevent miscalibration of SUV.
Research Support L. MacDonald has recieved research support from Philips Healthcare