P026] Dose and image quality evaluation in computed tomography with an anthropomorphic phantom

Abstract

Purpose Clinical PET/CT exams include a CT for attenuation correction and anatomical localization. The purpose of this study is to: • Evaluate if an anthropomorphic phantom is suitable to mimic clinical conditions regarding CT automated current modulation behaviour. • Describe the variation of the image noise as a function of reference mAs in the phantom, in order to predict noise level in clinical patients. Methods CT scans were performed with a 64-slice Somatom Definition AS CT mounted in a Biograph mCT TrueV PET/CT system, with 120 kV, 0.5 s rotation time, pitch 1 and automated current modulation. A sample of 94 patients scanned with our PET/CT clinical protocol (80 reference mAs) was reviewed. Patients similar to the “standard-man” were selected (N = 18, men with weigh 65–75 kg). The parameters of the CT exploration (effective mAs and CTDIvol) were registered. An anthropomorphic phantom (3D Torso Phantom, CIRS) was scanned with different reference mAs. Effective mAs and CTDIVol were recorded. Several circular regions of interest were drawn in lung, bone and soft tissue. Noise was quantified as the mean of standard deviation in Hounsfield Units (HU). Results Clinical CT for the “standard-man” was acquired with 49.5 ± 5.7 effective mAs and CTDIvol of 3.62 ± 0.4 mGy. When the CIRS phantom was scanned with 80 reference mAs, 45 effective mAs and a CTDIvol of 3.28 mGy were produced. One-sample t-test showed statistically significant differences between the patients sample and the phantom (p Noise measured in the phantom with 80 reference mAs was 41.9 HU in bone, 15.2 HU in lung and 22.6 HU in soft tissue. When reference mAs varied from 10 to 120 mAs, noise in lung, bone and soft tissue varied as a potential function of effective mAs with R2 greater than 0.94. Conclusions There is no concordance between the anthropomorphic phantom and the clinical studies. A function to model the relation between effective mAs and noise has been determined with a good fit.