Digital twin creation of a proton therapy treatment environment with hybrid LiDAR and RGB 3D camera

This study evaluates the feasibility of utilizing a commercial-grade 3D LiDAR/RGB camera, combined with 3D reconstruction software, to create an accurate and hyper-realistic digital model of a proton therapy treatment room. This reconstructed model aims to enhance collision avoidance strategies and improve 3D machine model accessibility. Utilization of a 3D LiDAR/RGB camera as a radiotherapy environment modeling has not been reported prior.

To create a comprehensive 3D model of a half-gantry proton therapy system, colored point cloud data was captured using the Matterport Pro 3 camera, a hybrid LiDAR and RGB imaging system with high-resolution capabilities (134.2 MP), a 360° horizontal field of view, and 20 mm accuracy at 10 m. A total of 117 acquisition points, distributed across the treatment room with three camera heights, ensured complete coverage and minimal occlusions. Scanning, completed in under 3 hours, was monitored in real time using the Matterport App. Post-scan processing involved denoising and converting the point cloud into 3D mesh structures using MeshLab, followed by refined pair-wise iterative closest point (ICP) alignments. Textures and materials were assigned to reflect real-world objects, and a ray-tracing engine simulated realistic lighting in Blender. Animations illustrating the kinematics of the treatment couch and gantry were simulated in Blender for enhanced visualization.

The scanning process achieved complete data capture without missing information, attributed to strategic oversampling during data acquisition. The mean root mean square error (RMSE) of the ICP registration was 0.008 m. The validation process confirmed that the dimensions of the treatment couch within the digital model closely matched the actual measurements with less than 2 cm deviation, indicating high accuracy. The resulting digital twin provided a photorealistic and immersive representation of the proton therapy treatment room, serving as a valuable digital asset for various applications.

It is possible to generate vendor-independent and highly accurate 3D models of the proton treatment room environment with a commercial grade LiDAR/RGB camera to expand future research opportunities and education endeavors.