Variations of Dose to the Lung During Computed Tomography (CT) Thorax Examinations: A Monte Carlo Study

This study determined the influence of patient individuality on lung organ doses for chest computed tomography (CT) examinations, viewed in the context of the recommendation of the ICRP 103. Within this current recommendation, a more individualized dose estimation is emphasized. The new ICRP 110 voxelized adult phantom was used and compared to calculation of lung doses for chest CT studies with identical scan parameters (120 kV, 135 mAs, 100 mm collimation, 1.5 pitch). For all patient images, the lung was contoured, and the scanning geometry was simulated using the Monte Carlo method. The lungs were completely included in the scan area. A user code was developed for the Monte Carlo package EGSnrc, which enables the simulation of a CT examination procedure and allows an efficient dose scoring within a patient geometry. All simulations were calculated with the same CT source model and calibrated to a realistic CTDIair value. Simulation values were grouped into 1 mSv classes. The organ dose classes fit well to a Gaussian distribution (adjusted correlation coefficient R2 = 0.95). The mean value of the fit was 10 mSv, with a standard deviation of 2 mSv. The variability was about ±30% with a minimum of 8 mSv and maximum of 13 mSv. The calculated lung dose of the ICRP adult female phantom was approximately 11 mSv and thus within the calculated standard deviation of the patient pool. The correlation between lung volume and dose was weak (adjusted correlation coefficient R2 = 0.33). Gender specific differences between the ICRP male and female phantoms were about 17%. In comparison, the differences between the female and a limited set of male patient studies were not statistically significant. Further, the relation between the HU values of CT scans and material/density necessary for the Monte Carlo simulations was investigated. It resulted that the simple but commonly employed relationship leads to significant deviations compared to definite materials in the ICRP phantoms. In conclusion, the present study demonstrates the expected variance in dose estimation due to the individual patient anatomy based on Monte Carlo simulations. The new ICRP phantoms properly represent the mean of the investigated patient pool. Nevertheless, a more individualized approach should be aimed for in the future. (author)