Room: Exhibit Hall | Forum 4
Purpose: Demonstrate an algorithm to prevent accidental target coverage errors created by inappropriate beam apertures.
Methods: Using the Eclipse Scripting API (Varian Medical Systems, Palo Alto, CA), target geometry and beam parameters were extracted for 116 clinical patient plans and 2 test plans. The clinical plans consisted of 66 VMAT, 21 IMRT, and 27 3D plans. The test plans contained induced errors to completely shadow or partially shadow the target with the aperture. 3D polygon models of the light field and collimator blocks were created which extended into the patient space. For each control point, overlaps between the models and the target geometry were calculated using the Robust Polygon Interference Detection (RAPID) algorithm. The results allowed for determination of three possible states of the target: inside, outside, and partially outside aperture.
Results: For all clinical cases, all targets were contained in the beams-eye view for each control point. The calculation time was 4.74 Â± 3.46 seconds per patient, and 19.72 Â± 14.99 ms per control point. For the test plans, the algorithm successfully detected the error of both the outside aperture and partially outside aperture states.
Conclusion: We have demonstrated a technique to automate error detection in aperture definitions to supplement traditional beams-eye view image review of the target.
Funding Support, Disclosures, and Conflict of Interest: Authors have received research funding from Varian Medical Systems.