Relative Error: An Approach for in vivo Characterization of Electromagnetic Tracking Errors and Confidence Intervals

摘要

The sensitivity of electromagnetic (EM) tracking to electromagnetic field distorters in a clinical environment is well known. Correction of the EM field distortion is challenging given the ill-posed nature of the problem and the typically sparse set of observations available in computing an inverse solution. Furthermore, interventional environments dynamically change during a procedure making pre-procedural calibration measurements difficult to apply. In this paper, we present the use of relative measurement errors derived from known geometric features of interventional instruments to detect reliable regions of operation. Relative error can be estimated from consecutive measurements of EM sensor locations without the need for additional information beyond the geometry of an interventional tool or in vivo phantom. We demonstrate the use of a calibration phantom that allows us to determine the statistical relationship between "relative" errors and "absolute" errors. Using this statistical relationship, we develop a method to compute an estimate of the absolute error, which provides real time information about measurement confidence.