Abstract
Recently, electromagnetic tracking for surgical procedures has gained popularity due to its small sensor size and the absence of line-of-sight restrictions. However, EM trackers are susceptible to measurement noise. Indeed, depending on the environment, measurement uncertainties may vary considerably. Therefore, it is important to characterise electromagnetic measurement systems when used in a fluoroscopy setting. The purpose of our study is to assess decoupled static electromagnetic measurement errors in position and orientation, without adding potential interference, in the presence of fluoroscopic imaging equipment.
Using an Aurora electromagnetic tracking system (Northern Digital, Waterloo, Canada), 5 degrees of freedom measurements were collected in a working space located midway between the source and the receiver of a flat-panel 3D fluoroscope (Innova 4100, GE Healthcare, Buc, France) emitting X-rays. In addition, to determine potential EM distortion from X-rays, electromagnetic measurement accuracies, as a function of position, were compared before, during, and after X-ray emissions. To decouple position and orientation errors, two scaffold devices were designed. Their centre was placed approximately at X = −50, Y = 0, and Z = −300 mm in the EM tracker's global coordinate system. First the positioning scaffold was used to assess the position and orientation measurement uncertainties as a function of position. Next, the orienting scaffold was used to assess the position and orientation measurement uncertainties as a function of orientation. Then, a least-squares method was employed to register the path position measurements to the known geometry of the scaffolds. As a result, the position accuracy was defined as the Euclidean distance between the registered and the ground truth positions. Finally, the orientation accuracy was defined as the difference between two direct angles: the angle between two measured consecutive paths, and the angle of the corresponding ground truth.
When translating the sensor using the positioning scaffold, the resulting position accuracy was characterised by a mean of 3.2 mm. Similarly, when rotating the sensor using the orienting scaffold, the resulting orientation accuracy was characterised by a mean of 1.7 deg. As for the “cross-displacement” errors, the orientation accuracy as a function of position had a mean of 1.8 deg. Likewise, the position as a function of orientation had a mean of 4.0 mm. Position and orientation accuracies – as a function of position, before, during, and after emission of X-rays – indicate that there was no significant interference by the presence of an X-ray beam on the EM measurements.
This work provides evidence that placing the EM system into X-ray beams does not affect EM measurement accuracies. Nevertheless, the fluoroscope itself significantly increases the EM measurement errors. Careful analysis of the EM measurement distribution errors suggests that associated uncertainties are predictable and preventable. In essence, EM tracking is promising for orthopedic procedures that may require the use of a fluoroscope.
