Increasing the Accuracy of Robotic Neurosurgical Procedures Through Robot Calibration

Abstract

One of the main concerns in neurosurgical procedures, besides safety and achieving sterile environments, is procedure accuracy. Although robots are well known for their reliability, absolute accuracy has always been an issue, especially for configurations with serially linked revolute joints. This issue can be addressed through various localisation and positioning strategies which significantly increase procedure accuracy but also increase overall procedure duration and can never achieve the full potential of the system. To address this issue robot calibration is performed to better fit the model used for positional calculations to the robot. In this article an experimental setup based on the neurosurgical robotic system RONNA is presented. A robot tool with highly precise sensors combined with a precisely measured reference phantom is used to perform measurements as well as for validation, and robot parameters are calibrated using a Sequential Quadratic Programming-based algorithm. A method for implementing the calibrated model into the RONNA workflow is presented and positioning accuracy achieved using the calibrated model is compared with a model obtained through the commercial calibration software RoboDK Validation results show positioning accuracies comparable to the commercially calibrated model, with additional improvements yet to be implemented into the model and calibration setup. Experimental testing shows a potential for decrease in localisation time and increase in overall procedure accuracy of neurosurgical procedures, which can be very beneficial, especially for more complex procedures.