Diagnostic CT organ dose estimation in whole-body PET/CT examinations–comparison of a patient-specific monte carlo approach and a computational phantom-based CT dosimetry tool
摘要
Studies evaluating the impact of advances in CT dosimetry tools on CT organ dose estimations are often limited to a comparison with TLD measurements in anthropomorphic phantoms or a comparison of different dosimetry tools using computational phantoms and CT examinations performed at radiology departments. This study evaluates organ dose estimations obtained using a patient-specific Monte Carlo simulation and a computational phantom-based dosimetry tool for whole-body PET/CT examinations. In addition, the correlation of organ doses with the size-specific dose estimate (SSDE) was investigated.
MethodsUsing the Monte Carlo software ImpactMC, patient-specific organ doses were simulated in 100 adult patients using whole-body CT scans acquired on a Siemens Biograph mCT Flow and a GE Discovery MI PET/CT. For each patient, organ doses were also estimated using the computational phantom-based dosimetry tool NCICT. Absolute and normalised to CTDIvol organ doses and percentage dose differences were assessed for CT acquisitions performed with tube current modulation (TCM). Statistical and regression analysis was performed to evaluate dose differences, their correlation with patient characteristics and the relationship with SSDE.
ResultsThe average percentage difference of NCICT to ImpactMC organ doses across all organs and BMI categories for whole-body examinations performed with TCM was − 5% and − 22% for the Siemens and GE PET/CT, respectively. Strong variations are observed between patients. Depending on the organ of interest, NCICT under-or overestimates the organ dose. Nevertheless, depending on the PET/CT system, moderate to excellent agreement was found between organ doses estimated with NCICT and ImpactMC. No correlations were observed between the obtained organ dose differences and patient length (R2 < 0.1), while weak to no or moderate correlations were found with patient weight (0.2 < R2 < 0.6) and BMI (0.2 < R2 < 0.7). Very strong correlations (R2 > 0.9) were observed between the estimated organ doses and SSDE.
ConclusionCompared to the patient-specific Monte Carlo CT dosimetry software ImpactMC, the computational phantom-based dosimetry tool NCICT could provide organ dose estimates within ± 22% for whole-body CT scans acquired with TCM. If better accuracies are required, patient-specific Monte Carlo simulations are recommended. Depending on the organ of interest and the specific CT scanner, SSDE may be a good first estimate of the organ dose.