Investigation of Image Extrapolation Methods to Improve Accuracy of CT Dose Calculations Based On Patient CT Images
Q Liang*, L DeWerd, University of WI-Madison/ADCL, Madison, WIWE-A-218-11 Wednesday 8:00:00 AM - 9:55:00 AM Room: 218
Purpose: The radiation dose received by patients from computed tomography (CT) scans has become a major public concern. A promising approach for individual CT dose determination is to use Monte Carlo (MC) simulations to calculate the dose to a patient based on their CT images. To accurately calculate the organ dose in the prescribed scan region, a whole body image is needed to properly simulate scatter effects. However, whole body images are not typically acquired for clinical exams. To address this issue, image extrapolation beyond the scan region was used, and its impact on dose calculation accuracy was investigated.
Methods: An EGSnrc-based MC tool, GMctdospp, was used to perform dose calculations. Dose distributions within the scan region, based on images with and without the region beyond the scan region (scatter region), were calculated. A mathematical cylindrical water phantom and a Rando phantom were used to demonstrate the effects of the scatter. For the Rando phantom, whole body and chest DICOM images were used to perform dose calculations. The doses calculated for these two sets of images were compared. Linear, polynomial, and nearest neighbor extrapolations of the slices on the edge were performed to determine the size of extrapolation region needed to provide enough scatter. The calculated dose based on chest images with extrapolated images were compared against the calculated dose based on the whole body images.
Results: Without the image extrapolation, an average of 10% dose underestimation was found on the edge of scan region for the water cylinder and up to 30% dose underestimation for the Rando phantom. With image extrapolation, the dose underestimation was negligible for these two cases.
Conclusions: Extrapolation of the images on the edge of scan region provides an effective way to improve the accuracy of determining patient-specific CT dose based on the CT images.