研究目的
To enable a temporary setup of a respiratory monitoring system in radiotherapy by developing a fast method to automatically calibrate the geometrical position using a near-infrared camera.
研究成果
The developed automatic calibration method for a near-infrared camera achieved high spatial accuracy (0.02 mm) and angular accuracy (1.6°), with an average random error of 0.46 mm per frame. This enables quick and temporary setups for respiratory monitoring in radiotherapy, facilitating frequent recalibrations and flexible camera positioning without permanent installation.
研究不足
The study used a specific phantom size (50 mm markers), which may not be optimal for all clinical scenarios. The method assumes alignment with treatment room lasers and may have dependencies on camera angle and measurement time in unoptimized cases. The accuracy could be affected by fabrication errors of the calibration markers.
1：Experimental Design and Method Selection:
A calibration phantom with seven markers was used, with one at the isocenter and six at 50 mm distances along orthogonal axes. A Kinect v2 NIR camera was set at ten different angles around the phantom with a fixed tilt of approximately 45°. Data were collected at 30 fps for 60 seconds, and subsets for 1, 3, 10, 30, and 60 seconds were analyzed. An initial conversion matrix was derived from the camera coordinates, followed by optimization using the least squares method to minimize residuals in the isocenter coordinate system.
2：5°. Data were collected at 30 fps for 60 seconds, and subsets for 1, 3, 10, 30, and 60 seconds were analyzed. An initial conversion matrix was derived from the camera coordinates, followed by optimization using the least squares method to minimize residuals in the isocenter coordinate system.
Sample Selection and Data Sources:
2. Sample Selection and Data Sources: The calibration phantom with markers was placed on a treatment couch aligned with lasers. Data were acquired from the camera's coordinate system.
3：List of Experimental Equipment and Materials:
Kinect v2 camera (Microsoft Corp.), calibration phantom with seven markers, linear accelerator treatment vault, Marker Tracker software (Mizuho Information and Research Institute, Inc.), and a spreadsheet for data processing.
4：Experimental Procedures and Operational Workflow:
The camera was positioned at various angles, data were recorded, and coordinates were exported to a spreadsheet. Initial parameters were estimated analytically, then optimized to convert camera coordinates to isocenter coordinates.
5：Data Analysis Methods:
Statistical analysis included calculating standard deviations, spatial accuracy, angular accuracy, and random errors. Fitting functions were used to model errors as functions of angle.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
