A Novel In-House 3D Surface Scanning System for Collision Detection Application In Radiotherapy
Abstract
Purpose
A novel, low-cost 3D surface scanning system was developed using a custom modification of the laser-profiling technique. Proof of concept was demonstrated through 3D simulation in Blender 3D software. A real-world prototype was then implemented using a consumer-grade camera and the existing laser in the CT-simulation (CT-Sim) room.
Methods
A single laser line source and multiple cameras were used, with the laser and cameras intentionally separated—unlike commercial laser profilers. The cameras were placed at optimized angles and distances to enhance visibility and reduce occlusion. The cameras captured the laser projected onto the object being scanned to extract 3D surface information. For simulation, Blender 3D was used to build a digital twin of the setup and emulate the laser within the CT‑Sim to image a human‑like digital phantom. For the real-world prototype implementation, a foam phantom was placed on the CT‑Sim table, and a tripod‑mounted camera recorded the laser projection during table motion at 120 fps in 4K. Intrinsic and extrinsic camera calibrations were performed using a checkerboard pattern. For validation, the 3D mesh of digital phantom and a 3D scan of the physical phantom, obtained with a commercial 3D scanner, served as ground-truth. The image‑processing pipeline was implemented in Python, and 3D surface analysis was performed in CloudCompare software.
Results
Simulation results showed an excellent agreement between the reconstructed 3D point-cloud and ground-truth mesh of the digital phantom with mean±SD of -0.1±0.8mm. For the physical phantom, the reconstructed 3D surface agreed with the ground truth within -0.48 ± 1.2 mm.
Conclusion
We have successfully demonstrated a cost-effective and accurate 3D surface scanning system using the laser source in the CT-Sim room and consumer-grade camera. This system can be integrated into the current clinical simulation workflow to produce the 3D surface of patients and immobilization devices for collision detection purposes.