Abstract
3195
Introduction: (1) To optimize CT technique so as to harmonize dose across different PET/CT scanners, each with vendor-specific tube current modulation. (2) To assess the applicability of CT dose estimation methods over the skull base-to-thigh and vertex-to-toe scan ranges.
Methods: Oncological PET/CT studies were acquired on two scanner systems (Siemens Biograph mCT and GE DVCT) and two relevant scan ranges (arms-up skull base-to-thigh and arms-down vertex-to-toe). Both systems used tube current modulation but the manufacturer’s implementations and parameters were not directly comparable. The mCT used CAREDOSE 4D, 80 eff mAs ref, 0.8 pitch, 0.5 sec / rotation, 120 kV. The DVCT initially used AutomA, 30-200 mA, Noise Index = 20, 0.985 pitch, 0.5 sec / rotation, 120 kV. As part of the present study, the DVCT parameters were adjusted so as to better match the CTDIvol of the mCT over a range of patient body sizes. Effective dose was estimated using two methods. Dose method 1 multiplied the dose length product (DLP from the scanner’s patient-specific CT dose report) by 0.013 mSv mGy-1 cm-1 (Inoue et al, 2015). Dose method 2 extracted CTDIvol data for each CT slice, determined an average DLP for various anatomical regions (head, neck, chest, abdomen/pelvis, proximal thigh, knee and distal leg), multiplied these regional DLP data by an appropriate scale factor (0.0021, 0.0059, 0.014, 0.015, 0.011, 0.0004, 0.0002 mSv mGy-1 cm-1 from AAPM Report No. 96, 2008 and Saltybaeva et al, 2014) and summed to estimate effective dose. Data from 200 PET/CT patient studies were analyzed.
Results: For the skull base-to-thigh scan range, mean CTDIvol was 6.4 ± 2.3 mGy (mCT, n=50) and 7.5 ± 0.8 mGy (DVCT original protocol, n=50). Not only was the average CTDIvol different between scanners butthe trend with body mass index (BMI) was also different. The DVCT protocol was adjusted to use both AutomA (z direction modulation) and SmartmA (angular modulation), 30-300 mA, Noise Index = 29, resulting in a mean CTDIvol = 6.2 ± 2.5 mGy (DVCT new protocol, n=50). With the adjusted protocol, mean CTDIvol was very similar between the two scanner systems and the dependence with BMI was highly comparable. Effective dose was also similar between scanners: 7.6 ± 2.6 mSv (mCT dose method 1) vs 7.5 ± 3.2 mSv (DVCT dose method 1). Note BMI and scan length in these two patient groups were similar: 27.8 ± 6.1 kg/m2 (mCT) vs 26.8 ± 5.7 kg/m2 (DVCT), 92.3 ± 6.8 cm (mCT) vs 92.9 ± 6.9 cm (DVCT). The two dose estimation methods were not greatly different over the skull base-to-thigh scan range: 7.6 ± 2.6 mSv (mCT dose method 1) vs 7.4 ± 2.7 mSv (mCT dose method 2). However, in a separate population (n=50) involving the vertex-to-toe scan range, dose methods 1 and 2 differed significantly (paired t-test, P << 0.05): 11.1 ± 2.7 mSv (mCT dose method 1) vs 10.0 ± 2.7 mSv (mCT dose method 2). Effective dose estimated using method 1, based on the average DLP over the entire scan, was higher than the region-specific approach (method 2) by ~12 %. Using dose method 2, effective dose / DLP was 0.0125 ± 0.0006 mSv mGy-1 cm-1 for skull base-to-thigh and 0.0116 ± 0.0005 mSv mGy-1 cm-1 for vertex-to-toe CT.
Conclusions: CTDIvol can be approximately matched across different scanner systems, each with vendor-specific tube current modulation techniques. Two different methods of estimating CT effective dose were in close agreement for skull base-to-thigh studies but differed significantly for vertex-to-toe studies. This difference is likely related to a wide variation of tube current and CTDIvol over the vertex-to-toe range, often deviating substantially from the average CTDIvol used for simple dose estimation (method 1). Improved estimation of effective dose can be obtained by using a DLP scale factor appropriate for the vertex-to-toe range or by specifically considering the CTDIvol profile across all slices (method 2).
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.