Dose Measurement and Estimation Methods in An Elliptical Body Phantom for a Conebeam CT System
J-H Choi*, D Constantin, G Nelson, R Fahrig, Stanford University, Stanford, CASU-D-103-7 Sunday 2:05PM - 3:00PM Room: 103
Purpose: To propose new metrics to estimate a mean dose of an automatic exposure control-enabled angiographic C-arm system over a noncircular large body-shaped phantom based on multi-point dose measurements.
Methods: Dose was measured at 9 points in 2 central subregions (C1, C2) and 14 points in 2 peripheral subregions (P1, P2) of the body phantom using a small 0.6cc ion chamber (IC) while operating the system at 16 different combinations of tube voltage, detector dose request, and vertical collimation. In order to acquire complete 2D dose profiles in an axial direction, we carried out Monte Carlo (MC) simulations. After validating the MC model by comparing it to chamber values, the mean dose from MC simulations was used as a ground truth for our mean dose metrics. Mean dose was estimated in 3 ways: 1) Area ratio for each point weights the contribution of the point. 2) Point dose surface fitting method using biharmonic interpolation model. 3) The acquired MC dose profile-based MC template method. We investigated how each method's performance varies as a function of the number of measured data points (7~23 points).
Results: The error of MC compared to chamber readings was 0.9mGy (+/-0.03 mGy) per point. The relative errors of 1), 2), and 3) methods with 23 IC points in comparison with the MC mean dose were 0.6, 3.37, and 1.9%, respectively. Method 1 performed best for 6 different cases of number of points measured. However, its performance fluctuated compared to methods 2 and 3. Method 3 remained within 3% error with 23~8 points and showed the most stable performance. Method 2 performed worst with 23~11 points, with constant error of ~ 5%.
Conclusion: The 3 metrics estimated a mean dose accumulated in a body phantom with about 5% relative error using only 11 points.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by National Institutes of Health (NIH SIG S10 RR026714-01), by Siemens Medical Solutions, AX, and by the Lucas Foundation. There is no conflict of interest to disclose.