Electromagnetic instrument tracking in computer-assisted interventions

Abstract

Accurate localization of medical instruments with respect to patient anatomy is an integral part of computer-assisted interventions. Electromagnetic tracking is the primary choice, particularly for minimally invasive procedures subject to visual tracking occlusion.

However, electromagnetic measurements are susceptible to field distortion by magnetic and electrically-conductive objects, including medical imaging devices, equipment, and instruments. Furthermore, in the majority of applications it is infeasible to mount electromagnetic sensors directly on the distal end of the instruments (tool tip); thus, the sensors are placed at the proximal end to indirectly observe the tool tip. In this case, measurement orientation error as well as instrument deflection can lead to significant tool tip tracking error. In clinical environments, the resulting tracking error can be in the order of a few centimeters. Unless compensated for, this error may severely compromise the outcome of the procedure.

This research is aimed to alleviate the aforementioned electromagnetic instrument tracking limitations. We present advanced methods for dynamic field distortion compensation and deflection estimation. Specifically, we propose to integrate the motion model of the tracked instrument with the sensor observations and apply a simultaneous localization and mapping algorithm to both accurately estimate the pose of the instrument and create a map of the field distortion in real-time for dynamic calibration. Furthermore, for flexible instruments, such as needles, we propose deflection estimation methods that anticipate instrument bending during insertion into deformable tissue. These methods rely on statistical sensor fusion techniques to combine kinematic deflection models with the position measurements in order to reduce the estimation error caused by uncertainties inherent in electromagnetic measurements and in the quantification of deflection model parameters.

For field distortion compensation, we conducted experiments in the presence of ferromagnetic and conductive field distorting objects, then compared our results to alternative statistical fusion methods. For instrument deflection estimation, we evaluated the performance of our multi sensor fusion approach during simulated needle insertions and brachytherapy phantom experiments. We also performed sensitivity analysis for plausible intra-operative conditions. With reduced tracking error and improved deflection estimation, our approach is promising for reliable electromagnetic navigation in various image-guided medical interventions.