(Wednesday, 8/1/2018) 10:00 AM - 10:30 AM
Room: Exhibit Hall | Forum 5
Purpose: To investigate the feasibility of monitoring MV beam’s eye view (BEV) pancreas position during VMAT breath-hold treatments using a stereo source-detector pair and digital tomosynthesis (DTS).
Methods: Simulation studies were conducted to determine whether displacements of up to 7 mm, consistent with drifts during a breath-hold, could be detected with high accuracy. The simulated rotating gantry configuration comprised a stereo kV source-detector pair placed symmetrically about the MV treatment head. Known shifts were applied to the pancreas and other organs in XCAT phantom volumes, and realistic x-ray projections containing kV and MV scatter and other noise sources were generated using the iTools research software (Varian, Palo Alto CA). DTS images were then reconstructed from the projections using the FDK algorithm, and a template matching algorithm was applied to the DTS volumes to estimate pancreas center-of-mass (COM) in the plane orthogonal to the BEV. The accuracies of these estimated BEV COM positions were investigated as a function of imager separation angle α and gantry rotation angle β per DTS acquisition. Similar studies were conducted using clinical abdominal CT images.
Results: As expected, BEV localization accuracy increased with larger β. Smaller α tended to increase accuracy particularly for values of β less than 30°. The clinical and XCAT results were consistent with each other with better than 1 mm shift estimation accuracies generally achieved for all values of β greater than 30° and for all values of α studied ranging from 45° to 90°.
Conclusion: DTS using a stereo source-detector pair may provide for high quality intra-fraction BEV pancreas localization with temporal resolution still being adequate for a breath-hold scenario. For example, if using a ring gantry with a 6 RPM rotation rate, 30° gantry rotation angles occur every 0.83 seconds. Future work will include investigating more sophisticated reconstruction algorithms.
Funding Support, Disclosures, and Conflict of Interest: Authors are employed by Varian Medical Systems
Keywords
Taxonomy
Not Applicable / None Entered.
Contact Email
