Endoscope calibration with real - time augmented reality video overlay : an IGSTK implementation

curacies within navigation software was investigated. Novel hybrid tracker configurations included use of the Window FG with infrared and/or video-based trackers. Results The fiducial registration error (FRE) between fiducials on the vertical phantom and divot targets was (1.12 ± 0.39), (1.19 ± 0.59), and (1.30 ± 0.91) mm in (x, y, z) directions for the Window FG, respectively, and (1.49 ± 0.63), (0.96 ± 0.44), (2.33 ± 1.70) mm for the conventional Aurora. As shown in Fig. 2a, the TRE measured from 10 repeat localizations was (1.43 ± 0.77), (1.29 ± 0.89), and (1.19 ± 0.71) mm in (x, y, z) directions for the Window FG, respectively, and (1.28 ± 0.51), (0.77 ± 0.48), (1.49 ± 0.86) mm for the conventional Aurora. A slight increase in TRE was observed for the Window FG near one FOV boundary (-y), and the Window FG exhibited superior TRE at greater depth (z). Overall, the magnitude and trend in TRE was similar for each tracker, with the Window FG exhibiting slightly improved accuracy and FOV. There was no change in TRE measured with and without a carbon-fiber plate. The Window FG aperture was sufficiently large for unattenuated PA fluoroscopy at any level of magnification for the 30 9 30 cm detector. CBCT images acquired without and with the Window FG under the OR table are shown in Fig. 2b. Although the FG enclosure was visible in oblique projections, and only a simple (Gaussian extrapolation) truncation correction was applied, truncation artifacts arising from the Window FG were minimal. A subtle increase in streak artifact is detectable, but the increase in image noise was minimal (*±2–5 HU) at the level of quantum noise. Expert surgeon feedback suggests a potentially important role for the tracker-in-table concept. These include clear logistical advantages and x-ray compatibility compared to a conventional over-table mount, and further effort is underway to realize the approach in clinically available operating tables. Conclusion The Window FG prototype exhibited comparable or superior TRE to the conventional Aurora and increased FOV (*15 cm greater depth). The open aperture and hollow frame yielded excellent x-ray compatibility for PA fluoroscopy and CBCT. Truncation artifacts from the tracker were minimal. Incorporation of the Window FG directly within the operating table is an enticing concept with improved logistics, workflow, and ‘‘invisible’’ integration of real-time tracking. The TRE and FOV are sufficient for a broad scope of applications, and the primary limitation remains susceptibility to EM field distortion. Hybrid tracking configurations exploiting the distinct strengths of EM and infrared/optical trackers are explored in future work.