191 / 236
Contouring pertinent anatomy allows intuitive display of
anatomical relationships during surgery. We recognized
the need for further interface development and aimed to
prepare for clinical introduction of this technology by
rigorously testing the system in a preclinical trial. To
truly identify the barriers to clinical implementation, we
planned an intensive operative exercise in a realistic
environment. We elected to exclude inexperienced sur-
geons or trainees as subjects to avoid results that may
not necessarily reflect the true clinical benefits and
costs. We restricted participants to fellowship-trained
otolaryngologists or neurosurgeons who regularly per-
form endoscopic skull base surgery. Multi-institutional
recruitment was required to achieve an experienced
cohort. There was a significant time commitment
(around 5 hours) from each participant, and the wet lab-
oratory was designed to replicate the operating room
(OR) environment as closely as possible.
Surgeons performed surgical tasks with the LIVE-
IGS system and also in a conventional manner so that
paired data could be collected. We focused on qualitative
feedback to identify interface design issues and target
potential improvements. This included visual display
settings, auditory alerts, and other ergonomic factors.
We intended to update the system in response to feed-
back throughout the trial. Task workload was assessed
to identify whether this technology altered the demands
on the surgeon.
MATERIALS AND METHODS
Seven otolaryngology (n
5
5) and neurosurgery (n
5
2)
skull base surgeons from five institutions participated in a
cadaver dissection trial. Prior to dissection, each head under-
went computed tomographic (CT) scanning followed by critical
structure manual segmentation using ITK-SNAP 2.0 software.
9
Structures contoured included the carotid arteries, optic nerves,
pituitary gland, dura, and orbits. Alert zones of approximately
2 to 3 mm were manually mapped around the carotid arteries
and the dura (Fig. 1). This process was undertaken by the
investigating surgeons and took approximately 60 minutes per
case. In keeping with clinical protocols, we initially performed
magnetic resonance imaging, but the quality of the scans was
poor on cadaver specimens and offered no advantage over CT
for the purposes of this study. CT angiography could not be per-
formed on cadavers.
The initial surgical approach was performed by the inves-
tigators before the subjects started the study task (clivus abla-
tion). Ethmoidectomy, posterior septectomy, and a wide, unified
sphenoidotomy were performed, and then the heads were reim-
aged with a surgical cone-beam CT (CBCT) system.
10
Deforma-
ble registration allowed the contours delineated from the
preoperative CT to be registered to the intraoperative CBCT
imaging.
11
Optical IGS reflective markers were attached to the head,
the 0 endoscope (Hopkins II telescope and IMAGE1 camera;
Karl Storz, Tuttlingen, Germany), and the drill (M4 hand-piece;
Medtronic, Jacksonville, FL). Registration of the head to the
imaging data was then undertaken with an optical tracking sys-
tem (Polaris; NDI, Waterloo, Ontario, Canada). Fiducial
Fig. 1. Manual segmentation of cross-sectional computed tomographic scan images was used to create a three-dimensional virtual view of
anatomical structures. A
5
anterior; I
5
inferior; L
5
left; P
5
posterior; R
5
right; S
5
superior.
Laryngoscope 124: April 2014
Dixon et al.: Real-Time Navigation for Endoscopic Surgery
169