Image registration in a tomographic ultrasound system: Comparison between camera-tracking and image-based motion compensation

We present a water tank-based tomographic ultrasound system for 3D ultrasound imaging. To date, the clinical focus of this setup has been limb imaging, with a particular emphasis on its application to prosthetic socket design. When performing scans on human subjects, natural patient movement presents a significant challenge as it often leads to image distortion, which inhibits accurate discernment of anatomical structure. In this paper, we compare two motion compensation methods that were created to compensate for limb motion: a camera-based motion tracking method and an image-based registration method. We show that camera imagery can effectively track and compensate for limb motion without requiring spatially overlapping images, nor a bundle adjustment stage to correct for drift. This method has the potential for rapid and robust stitching of circumferentially-collected ultrasound images into a volumetric dataset.