A framework for calibration of electromagnetic surgical navigation system

In this paper, we present a framework of calibrating an electromagnetic tracker (Northern Digital's Aurora) using an accurate optical tracker (the Optotrak system, also from Northern Digital). First, registration methods for these two navigation systems are developed. Sub millimeter accuracy registration is achieved for both cases. We also address the latency between the two different trackers. The registration accuracy for dynamic acquired data is greatly improved after we compensate for the tracker latency. In our calibration approach, we sample the measurement field of the Aurora and compute the position and orientation error using the Optotrak measurements and previously computed registration results as "ground truth". Then we approximate the error field using Bernstein polynomials. Another comparative technique we use is to decompose the error space using KD tree, and then approximate each atomic cell with local interpolation. Experimental results show significant improvement in tracking accuracy for both position and orientation. Finally we discuss our future directions.