Ultrasound-Based Visual Servoing for Out-of-Plane Longitudinal Needle Tracking in Robot-Aided Percutaneous Nephrolithotomy

摘要

Percutaneous nephrolithotomy (PCNL) is a minimally invasive procedure to remove large renal calculi through a small incision in the patient’s back. Ultrasound (US) imaging is commonly used to guide the needle to the kidney during this procedure. However, it requires an advanced level of dexterity to coordinate the US probe and the needle to keep the needle visible in the images at all times. Failure to maintain needle-probe alignment can result in inadvertent injury, bleeding, and other complications. The use of robotic assistance can alleviate the surgeon’s cognitive workload by enabling autonomous positioning of the US probe and accurate needle tracking. This paper presents a new US-guided visual servoing (VS) algorithm for needle tracking using longitudinal US images of a needle subjected to out-of-plane motion. The ultrasound probe can move in 4 degrees-of-freedom (DOF), that is, two translations and one rotation in the imaging plane, and one rotation out of the imaging plane. Unlike previously reported VS algorithms, 4-DOF tracking is achieved using only 2D-US images and without any additional position sensor or prior knowledge of the needle trajectory. The algorithm is validated extensively in three different experimental scenarios using a water tank, a tissue phantom, and ex-vivo porcine tissue. Results obtained from several trials confirm the algorithm’s ability to track the needle and maintain needle-probe alignment with an average error of 1.5 mm, despite an out-of-plane average needle deflection of 7 mm along a 60 mm insertion depth.