Reproducibility of quantifying tracer uptake with PET/CT for evaluation of response to therapy

We present a study of the variability in quantifying tracer uptake in PET/CT, which is important for evaluation of response to therapy. Variability unrelated to response must be assessed to determine the threshold change required for declaring patient response to therapy. This study assessed systemic variability by repeatedly imaging a phantom filled with long half-life Germanium-68 in a solid epoxy matrix. Circular Region Of Interest (ROI) analyses with 10-mm diameters measured maximum and average activities of 10- to 37-mm diameter spheres. Three sets of repeated scans were acquired using a GE DSTE-16 PET/CT: (1) with a stationary phantom, (2) with sequential axial shifts of 0.5-mm between scans, and (3) with random repositioning between scans, also called 'coffee-break' repeat scans. For the stationary phantom scans the standard deviations for mean and maximum absolute Recovery Coefficients (RC) ranged from 0.9% to 4.3% depending on the methods of acquisition and reconstruction. In all cases similar results were observed for both analytic and iterative reconstruction methods. All standard deviations for maximum and average values for 2D acquisitions were significantly higher than those calculated from 3D acquisitions with 7-mm smoothing (p les 0.005). For the sequential axial shifted scans activity measurements did not demonstrate a systemic change corresponding to detector spacing. For the random repositioning scans, the standard deviations for maximum and average RCs were 2.9% and 2.4% averaged over all sphere diameters, compared to 2.4% and 1.4% for the stationary phantom with the same acquisition and reconstruction. Random axial repositioning scans acquired via a Siemens Biograph-16 Hi-REZ yielded similar results after the standard deviations were adjusted to compensate for differences in total counts acquired. These results suggest it is possible to account for the sources of variability in serial PET scans that are unrelated to response to therapy.