Electromagnetic Tracking System Assessment and Disturbance Compensation

OData support
Supervisor:
Dr. Haidegger Tamás Péter
Department of Control Engineering and Information Technology

Nowadays, electromagnetic tracking is considered to be a promising surgical navigation technology, though its applicability is highly limited due to the susceptibility to proximate metallic objects and electrical equipment. These outer impacts may cause distortion of the magnetic field, resulting in inaccurate measurements.

The main objective of my thesis was to investigate thoroughly the nature of measurement errors that characterize electromagnetic tracking systems (especially in the case of an NDI Aurora system). Based on the experiences, I developed a relatively flexible, cost–effective, fast and computationally efficient calibration protocol for assessment of magnetic field distortion.

After studying some available solutions to error correction being presented in the literature, I designed a special, cubic–shaped calibration phantom, made of acrylic glass, which enables the user to accomplish fast and systematic data acquisition within a certain range of the entire working volume. Complying with the numeric constraints defined by the calibration tool, appropriate spatial transformations are determined between corresponding points of sampled data and that of the reference frame. Applying these transformations to the data set, error correction vectors are computed that imply differences in position as well as in orientation at each grid point. Subsequently, we can generate the Delaunay tetrahedralization of the grid, which can be used as a good basis for posterior interpolation. As an alternative, we can also perform linear regression or nonlinear curve fitting on the calibration data, which results in an analytical model of the stationary distortion field. The extracted model provides a basis for online error compensation, for subsequent use of the system.

Having a major advantage over other concepts, the whole setup procedure can be executed in 10–15 minutes, and requires no particular a priori knowledge about the instrumentation. The graphical user interface of the application offers a practical way to configure the system. The implemented software is capable of displaying tracking data and animating it in 3D, while informing the user about the reliability of sensor transformations actually being reported.

Finally, outcomes of validation tests are presented, and the conclusion is drawn that tracking accuracy can be improved by a factor of 1.5 in position and 1.2 in orientation.

Downloads

Please sign in to download the files of this thesis.