Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

Virtual Collision Checks with a Laser Scanning and Depth Sensing Camera

B Posehn1*, H Clark2, (1) University Of British Columbia, Vancouver, BC, CA, (2) British Columbia Cancer Agency, Surrey, BC, CA


PO-GePV-M-206 (Sunday, 7/25/2021)   [Eastern Time (GMT-4)]

Purpose: Radiotherapy with non-coplanar beams typically require patients to attend a mock treatment ('collision check') wherein the motions of a radiotherapy plan are performed with the patient immobilized, but without radiation delivery. Treatment staff observe clearance throughout the simulation. Collision checks require LINAC time, staff time, and patient travel. Alternatively, plans and support structures can be conservatively designed to avoid the need for collision checks, but this can limit available beam orientations and patient setups (e.g., arms up) and result in sub-optimal plans. In this work we demonstrate that virtual simulation of collision checks are possible using a low-cost hand-held laser scanning and depth sensing (LiDAR) camera.

Methods: An inexpensive ($500 USD) camera was built using a Raspberry Pi, an Intel RealSense L515 LiDAR camera module, and a 3D-printed housing. Models of a General Electric CT scanner and Varian TrueBeam LINAC, including couches, were captured. Models were spatially registered, and the LINAC model was rigged for virtual motion. An anthropomorphic phantom placed on the CT couch was LiDAR scanned, registered to the base CT model, and a 3D mesh model was extracted. It was then virtually transferred to the LINAC. LINAC rotational motion was simulated via a Minkowski product and intersection tests were performed to identify intentionally-created collision scenarios.

Results: Spatial errors in the LIDAR-scanned surfaces were dependent on the material, ranging from the LINAC exterior (±0.7 mm) to glossy carbon fibre (±2.1 mm). In the absence of other positional uncertainties (e.g., support structures), couch-LINAC collisions due to couch translation could be predicted virtually within approximately 2.2 mm.

Conclusion: Virtual collision checks have the potential to reduce resource usage for plans with non-coplanar beams, atypical patient setups, and bulky support structures. By allowing radiotherapy planners to simulate collision checks as part of the planning stage, additional treatment options are possible.



    Not Applicable / None Entered.


    Not Applicable / None Entered.

    Contact Email