Motion-Aware Robotic 3D Ultrasound
Authors:
Zhongliang Jiang,
Hanyu Wang,
Zhenyu Li,
Matthias Grimm,
Mingchuan Zhou,
Ulrich Eck,
Sandra V. Brecht,
Tim C. Lueth,
Thomas Wendler,
Nassir Navab
Abstract:
Robotic three-dimensional (3D) ultrasound (US) imaging has been employed to overcome the drawbacks of traditional US examinations, such as high inter-operator variability and lack of repeatability. However, object movement remains a challenge as unexpected motion decreases the quality of the 3D compounding. Furthermore, attempted adjustment of objects, e.g., adjusting limbs to display the entire l…
▽ More
Robotic three-dimensional (3D) ultrasound (US) imaging has been employed to overcome the drawbacks of traditional US examinations, such as high inter-operator variability and lack of repeatability. However, object movement remains a challenge as unexpected motion decreases the quality of the 3D compounding. Furthermore, attempted adjustment of objects, e.g., adjusting limbs to display the entire limb artery tree, is not allowed for conventional robotic US systems. To address this challenge, we propose a vision-based robotic US system that can monitor the object's motion and automatically update the sweep trajectory to provide 3D compounded images of the target anatomy seamlessly. To achieve these functions, a depth camera is employed to extract the manually planned sweep trajectory after which the normal direction of the object is estimated using the extracted 3D trajectory. Subsequently, to monitor the movement and further compensate for this motion to accurately follow the trajectory, the position of firmly attached passive markers is tracked in real-time. Finally, a step-wise compounding was performed. The experiments on a gel phantom demonstrate that the system can resume a sweep when the object is not stationary during scanning.
△ Less
Submitted 13 July, 2021;
originally announced July 2021.
