Accuracy of a photogrammetry-based patient positioning and monitoring system for radiation therapy

A photogrammetry system designed to reduce simulator-to-treatment and treatment-to-treatment patient positioning errors has been developed. Two complete systems have been installed in our department: one in the simulator room and one in a treatment room. Each system consists of three charge-coupled device (CCD) cameras; a ring of infrared LEDs around the lens of each camera; and several small, circular, retroreflective markers that are applied to the patient. The markers reflect infrared light directly back to the cameras, producing a binary image of oval hot spots when the image is thresholded. The three-dimensional position of each marker is calculated by conventional photogrammetry methods. At simulation, marker positions are measured, then transferred to the treatment room system. The system may be used to actively position patients, and to passively monitor a patient's position and motion during treatment. Studies have focused on measuring the system's temporal stability, precision, and accuracy; on optimal positioning of markers and cameras; and on assessing the system's capability to reduce the positioning error. The repeatability of measuring a marker's position is <0.1 mm in each orthogonal direction. The accuracy is approximately 0.5 mm over a 40 X 40 X 40 cm3 field of view. The system drift over four hours is approximately +/-0.2 mm. The photogrammetry system has been used to actively position a lead BB, embedded within a head phantom, at the isocenter; repeatability was +/-0.3 mm, as determined radiographically. The system has also been used to passively monitor the positioning of several head and neck patients that were set up by a therapist; setup errors of up to 10 mm in each orthogonal direction were measured, as well as the motion of the patient during treatment.