Multiobjective Trajectory Tracking of a Flexible Tool During Robotic Percutaneous Nephrolithotomy

Abstract

Percutaneous Nephrolithotomy (PCNL) is the leading intervention for removing large or irregularly shaped kidney stones. It involves gaining access to the kidney through a small incision in the patient's back, through which a nephroscope is steered towards the stones. Despite decades of clinical prevalence, PCNL remains a complex procedure to learn and perform, sometimes requiring several attempts to gain kidney access, leading to a variety of complications. This letter proposes to use robotic assistance to steer a flexible nephroscope during PCNL to concurrently improve accuracy and reduce the risk of tissue damage. The nephroscope is modelled as a cantilever beam fixed to the robot's end-effector. Under the assumption that an optimal tooltip path exists, Non-dominated Sorting Genetic Algorithm-II is implemented to determine the end-effector position and orientation so that the tooltip follows the path while minimizing four objective functions, i.e., tissue compression, variations in the tool's strain energy, changes in end-effector position, and tooltip error. Data collected through experiments performed on ex-vivo porcine tissue show that the path tracking error was on average 2.03 mm. The results confirm the accuracy of the model in 2 dimensions and suggest that the multiobjective optimizer returned adequate solutions that minimized 4 different cost functions, altogether allowing the robot to effectively follow the predefined path.