Development of a Compact Radiographic Simulator Using Microsoft Kinect
M Ono1*, K Kozono1, M Aoki1, A Mizoguchi1, Y Kamikawa2, Y Umezu3, H Arimura4, F Toyofuku4, (1) Department of Health Sciences, Graduate School of Medical Sciences, Kyushu University, Fukuoka (2) Department of Health Sciences, School of Medicine, Kyushu University, Fukuoka (3) Division of Radiology, Department of Medical Technology, Kyushu University Hospital, Fukuoka (4) Department of Health Sciences, Faculty of Medical Sciences, Kyushu University, FukuokaSU-E-I-91 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
Purpose: Radiographic simulator system is useful for learning radiographic techniques and confirmation of positioning before x-ray irradiation. Conventional x-ray simulators have drawbacks in cost and size, and are only applicable to situations in which position of the object does not change. Therefore, we have developed a new radiographic simulator system using an infrared-ray based three-dimensional shape measurement device (Microsoft Kinect).
Methods: We made a computer program using OpenCV and OpenNI for processing of depth image data obtained from Kinect, and calculated the exact distance from Kinect to the object by calibration. The object was measured from various directions, and positional relationship between the x-ray tube and the object was obtained. X-ray projection images were calculated by projecting x-rays onto the mathematical three-dimensional CT data of a head phantom with almost the same size. The object was rotated from 0 degree (standard position) through 90 degrees in increments of 10 degrees, and the accuracy of the measured rotation angle values was evaluated. In order to improve the computational time, the projection image size was changed (512*512, 256*256, and 128*128).
Results: The x-ray simulation images corresponding to the radiographic images produced by using the x-ray tube were obtained. The three-dimensional position of the object was measured with good precision from 0 to 50 degrees, but above 50 degrees, measured position error increased with the increase of the rotation angle. The computational time and image size were 30, 12, and 7 seconds for 512*512, 256*256, and 128*128, respectively.
Conclusions: We could measure the three-dimensional position of the object using properly calibrated Kinect sensor, and obtained projection images at relatively high-speed using the three-dimensional CT data. It was suggested that this system can be used for obtaining simulated projection x-ray images before x-ray exposure by attaching this device onto an x-ray tube.