Performance characteristics of the 5‐ring GE Discovery MI PET/CT scanner using AAPM TG‐126 report

Abstract

AimTo report on the performance characteristics of the 5‐ring GE Discovery MI PET/CT systems using the AAPM TG‐126 report and compare these results to NEMA NU 2‐2012 where applicable.Materials and MethodsTG‐126 testing was performed on two GE 5‐Rings Discovery MI scanners. Tests performed included spatial resolution, PET/CT image‐registration accuracy, sensitivity, count rate performance, accuracy of corrections, image contrast, scatter/attenuation correction, and image uniformity. All acquired data were analyzed using scanner console or free software tools as described by TG‐126 and the results were then compared to published NEMA NU 2‐2012 values.ResultsBoth scanners gave similar resolution results for TG‐126 and NEMA NU 2‐2012 and were within manufacturer specifications. Image‐registration accuracy between PET and CT using our clinical protocol showed excellent results with values ≤1 mm. Sensitivity using TG‐126 was 19.43 cps/kBq while for NEMA the value was 20.73 cps/kBq. The peak noise‐equivalent counting rate was 2174 kcps at 63.1 kBq/mL and is not comparable to NEMA NU 2‐2012 due to differences in phantoms and methods used to measure and calculate this parameter. The accuracy of corrections for count losses for TG‐126 were expressed in SUV values and found to be within 10% of the expected SUV measurement of 1. Image contrast and scatter/attenuation correction using the TG‐126 method gave acceptable results. Image uniformity assessment resulted in values within the recommended ± 5% limits.ConclusionThese results show that the 5‐ring GE Discovery MI PET/CT scanner testing using TG‐126 is reproducible and has similar results to NEMA NU 2‐2012 tests where applicable. We hope these results start to form the basis to compare PET/CT systems using TG‐126.