Purpose: Non-coplanar treatment techniques such as dynamic trajectory radiotherapy (DTRT) have the potential to improve treatment plan quality in comparison to conventional volumetric modulated arc therapy. However, collision avoidance must be considered during treatment plan creation. In this work, we developed a detailed collision-prediction tool for gantry and table-top or table-base, considering possible six degree of freedom (DoF) set-up table correction.
Methods: A virtual linac model was created using blender, a free and open-source 3D computer graphics software toolset, that can be accessed via an integrated python API. DTRT plans described with XML files can be simulated on the virtual linac and tested for possible collisions between gantry and table for a given table position. If an additional tolerance table is provided, all combinations of maximal setup shifts given in the tolerance table are tested for potential collision interlocks. Collision prediction was verified experimentally by setting the gantry to a certain angle and then translating or rotating the table until a collision interlock occurred. The resulting table angles and positions were compared to the predicted values. This was repeated with additional 3° pitch and roll rotations in order to verify the model for setup correction.
Results: For collision due to table angle, the mean difference between predicted and measured table angles leading to collision interlock of the gantry with table-top or table-base were 2.1° and 4.2°, respectively, with 70% of the values within 2° difference. For collision due to table translations with and without setup correction, the mean difference between predicted and measured table positions between gantry and table-top was 0.7 cm with 97% of the values within 2 cm difference.
Conclusion: Using blender, a collision prediction framework between gantry and table capable of considering patient shifts with a 6 DoF table was successfully developed and verified with measurements.
Funding Support, Disclosures, and Conflict of Interest: This work was funded by the Swiss National Science Foundation
Not Applicable / None Entered.