Collision prediction software for radiotherapy treatments.

Area of Science:

• Medical Physics
• Radiotherapy Technology
• Image Processing

Background:

• Collision prediction is crucial in external beam radiotherapy to ensure patient safety and treatment efficacy.
• Current methods may not fully capture patient topography and immobilization devices, leading to potential issues.
• Surface imaging offers a promising avenue for improved collision detection during treatment simulation.

Purpose of the Study:

• To present a novel method for collision prediction in external beam radiotherapy utilizing surface imaging.
• To evaluate the clearance of patient treatment positions during simulation and enable necessary modifications.
• To enhance the safety and efficiency of radiotherapy treatment planning and delivery.

Main Methods:

• A Kinect camera and Skanect software were used for surface reconstruction of the patient and immobilization devices.
• The treatment isocenter was marked, and the surface point cloud was converted to cylindrical coordinates for analysis.
• A cylindrical model was employed to predict collisions during gantry rotation with the treatment couch at 0°.

Main Results:

• Experimental verification showed predicted collision angles closely matched room-recorded collisions (within 0.5° in most cases).
• The Kinect-Skanect surface reconstruction demonstrated an average accuracy of 2.2 mm compared to CT scans.
• The methodology proved to be fast and reliable for collision prediction.

Conclusions:

• The developed surface imaging methodology provides accurate and efficient collision predictions for radiotherapy.
• Comprehensive modeling of patient topography and immobilization devices aids in identifying potential collisions.
• Implementing this tool during treatment simulation can optimize patient positioning, improve safety, and streamline clinical workflows by minimizing replanning.