Ryan A. Bartholomew, MD; Haoyin Zhou, PhD; Alejandro Garcia, MD, Maud Boreel, Research trainee, ; Jeffrey P. Guenette, MD; Nir Ben-Shlomo, MD; Krish Suresh, MD; Daniel J Lee, MD; Jayender Jagadeesan, PhD; C. Eduardo Corrales, MD
Jayender Jagadeesan, PhD; C. Eduardo Corrales, MD
Research Category: Digital Health, Imaging, and Informatics
Purpose: We seek to validate a stereoscopic surface reconstruction approach to surgical navigation in the lateral skull base. To relate the exposed tissue surface to underlying anatomy, a 3D surface reconstruction is created with 3D endoscopy in real time and fused to preoperative CT and MRI images.
Methods: During multiple steps of a cadaveric translabyrinthine dissection, two 3D models are generated. One model is created by stitching video frames captured with 3D endoscopy using stereo matching. This 3D surface model is then aligned to a 3D segmented CT model with a novel feature-based simultaneous localization and mapping method. Models are created in 3D slicer and fused using artificial fiducials (1.2 mm screws). Registration accuracy is assessed by calculating fiducial target registration errors (TREs).
Results: At five timepoints during translabyrinthine surgery in two cadaveric specimens, we generated surface models using a few seconds of 3D endoscopic video and 3D segmented CT models. These models were fused with a mean TRE of 0.76 mm (standard deviation 0.44 mm)
Conclusions: Our preliminary findings suggest that these stereoscopic surface reconstructions may provide surgeons a navigation technique by which they can “see through” opaque bone in real time while drilling within millimeters of critical structures.”