Erratum: Wangerin et al. (2015)

Prior reports have suggested that delayed 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) oncology imaging can improve the contrast-to-noise ratio (CNR) for known lesions. Our goal was to estimaterealistic bounds for lesion detectability for static measurements within 1 to 4 hours between FDG injectionand image acquisition. Tumor and normal tissue kinetic model parameters were estimated from dynamic PETstudies of patients with early-stage breast cancer. These parameters were used to generate time-activitycurves (TACs) for up to 4 hours, for which we assumed both nonreversible and reversible models with differ-ent rates of FDG dephosphorylation (k4). For each pair of tumor and normal tissue TACs, 600 PET sinogramrealizations were generated, and images were reconstructed using the ordered subsets expectation maximi-zation reconstruction algorithm. Test statistics for each tumor and normal tissue region of interest were outputfrom the computer model observers and evaluated using a receiver operating characteristic analysis, with thecalculated area under the curve (AUC) providing a measure of lesion detectability. For the nonreversiblemodel (k4 = 0), the AUC increased in 11 of 23 (48%) patients for 1 to 2 hours after the current standardpostradiotracer injection imaging window of 1 hour. This improvement was driven by increased tumor/nor-mal tissue contrast before the impact of increased noise that resulted from radiotracer decay began to domi-nate the imaging signal. Ask4was increased from 0 to 0.01 min−1, the time of maximum detectabilityshifted earlier, due to decreasing FDG concentration in the tumor lowering the CNR. These results imply thatdelayed PET imaging may reveal inconspicuous lesions that otherwise would have gone undetected.