Calibration of Swept-Volume 3D Ultrasound

We present two different methods for calibrating 3D swept-volume ultrasound as a prerequisite for an ultrasound based intraoperative navigation system. Different solutions of the calibration problem for 2D ultrasound exist, but no approach has been published so far that takes advantage of 3D swept-volume ultrasound. In contrast to 3D freehand ultrasound no compounding has to be performed. Instead of points or planes as invariant calibration objects we use 3D structures, whose detection and registration is more robust. The utilization of a commercially available 3D ultrasound phantom provides high quality ultrasound images, enabling automatic detection resp. registration of the embedded structures as a preliminary study indicates. Because our calibration methods theoretically require only one resp. three 3D ultrasound acquisitions each lasting less than four seconds the duration of the whole calibration procedure depends primarily on the time needed for the registration task. The accuracy of the two calibration methods is compared by a simulation study, providing similar results and suggesting a significant improvement of accuracy if an optical instead of a magnetic tracking system is used. Additionally an experiment with a magnetic tracking system has been realised. These preliminary results show that fixed points can be located with 3.3 mm root mean square uncertainty when imaged from different viewpoints.