Room: Karl Dean Ballroom A1
Purpose: The increased use of non-coplanar arcs enhances the possibility of couch-gantry or patient-gantry collision, a safety concern. To prevent collisions from occurring, this work presents a collision avoidance computer program that customizes the prediction by using a patient’s segmentation and simulating gantry and 6 DOF couch motions according to the treatment plan.
Methods: A collision prediction algorithm based on oriented bounder boxes was deployed to the clinic to simulate individual treatment deliveries. The software module use virtual simulation to preview treatment scenarios using high-definition models of the linac and couch as well as the patient’s actual contours. The algorithm software is installed as an ESAPI script in Eclipse communicating patient and plan details to the virtual reality software that further measures and reports collisions or clearances for every arc\beam in the plan.
Results: To test the algorithm’s capability of predicting collisions, measurements were done on a Varian TrueBeamTM and the Varian PerfectPitchTM couch by setting random positions for angles\translations to measure the physical clearance. After about 300 measurements were made obtained measurements were compared to the values given by the computer program. Accuracy tests show an average error of 2.4 cm, with some potential “blind spots� that increases the error to 4.6 cm. Data analysis suggest the need of a 3.0 cm safety buffer to increase the collision prediction capabilities of the program.
Conclusion: This software is used 193 times since its clinical installation in January 2017, providing a clinical solution for dosimetrists, physicists and therapists to prevent injuries and equipment damage, while improving workflow and productivity by integrating seamlessly with the existing treatment planning software.
Not Applicable / None Entered.
Not Applicable / None Entered.