Room: AAPM ePoster Library
Purpose: Potential collisions between gantry and patient, couch, or various objects during treatment delivery poses a major challenge in radiotherapy and can result in patient injury or costly repairs. Current safety mechanisms provided by vendors can only mitigate collisions with the patient and don't consider collisions outside treatment zone. In this study, we present a novel method to predict in-room collisions by constructing an accurate 3D map of the entire treatment room using a light detection and ranging (LiDAR) technique.
Methods: 2D mapping was achieved using a laser range scanner (RPLIDAR A2, SLAMTEC). The device has a range of 12 meters and runs at frequency of 5Hz with angular resolution of 0.45 and distance resolution of 0.2cm. The device performs a 360 degree scan at a sampling rate of 8000/sec. Multiple 2D scans were performed at various locations and combined using simultaneous localization and mapping algorithm (SLAM) available in Matlab to generate high resolution 2D maps. The 2D maps were then combined to construct a point cloud representing the 3D map of the entire room.
Results: The 2D scans showed an accurate representation of the room floorplan. The room dimensions extracted from the 2D scan agreed with the physical dimensions (L=6.73m, W=6.77m). The accuracy was ±0.2cm for short range (15cm < distance <500cm) and ±1 cm for long range scans (distance>500cm). The accuracy was improved by combining multiple 2D scans taken at different locations using the SLAM algorithm. The mesh representation of a partial 3D scan showed an accurate depiction of all the components inside the treatment room.
Conclusion: A highly detailed and accurate 3D map of the treatment room was achieved using a low-cost LiDAR. This information is then used to verify the trajectory clearance of the gantry and couch to avoid collisions.
Computer Vision, Data Acquisition, Radiation Risk
TH- External Beam- Photons: Development (new technology and techniques)