Content uploaded by Alon Grinshpoon
Author content
All content in this area was uploaded by Alon Grinshpoon on May 12, 2020
Content may be subject to copyright.
3:54 PM Abstract No. 29
Augmented reality guidance for cerebral
angiography
G. Loeb
1
, S. Sadri
1
, A. Grinshpoon
1
, J. Carroll
2
,
C. Cooper
2
, C. Elvezio
1
, S. Mutasa
2
, G. Mandigo
2
,
S. Lavine
2
, J. Weintraub
2
, A. Einstein
2
, S. Feiner
1
,
P. Meyers
2
;
1
Columbia University, New York, NY;
2
Columbia University/New York Presbyterian Hospital,
New York, NY
Purpose: Augmented reality (AR) holds great potential for IR by
integrating virtual 3D anatomic models into the real world.
1
In this
pilot study, we developed an AR guidance system for cerebral
angiography, evaluated its impact on radiation, contrast, and fluo-
roscopy time, and assessed physician response.
Materials: In this prospective study, 9 patients with CT or MR
imaging of the aorta underwent diagnostic neuroangiography with
AR guidance from June to August 2017. Before each procedure,
segmentation software was used to create a 3D model of the pa-
tient’s aortic arch including carotid and vertebral arteries. The
model was deployed to HoloLens (Microsoft, Redmond, WA), a
stereoscopic optical see-through AR head-worn display. Using the
AR user interface we developed, physicians manipulated a virtual
3D model intraoperatively via voice commands, gaze, and gestures
while maintaining sterility. In total, 6 physicians completed 14
postoperative questionnaires assessing the system. 18 case-
matched retrospective controls were identified by screening for
age, aorta imaging, cone-beam CT, indication, physician, and OR.
Results: All 9 patients underwent diagnostic neuroangiography
per standard protocol with AR guidance without complication.
Mean kerma-area product 3150 μGym
2
(SD 2284), skin-absorbed
dose 283 mGy (SD 192), contrast volume 119 mL (SD 35), and
fluoroscopy time 10 min (SD 4) were below reference values for
diagnostic neuroangiography.
23
There was a non-significant
reduction in kerma-area product, skin-absorbed dose, and fluo-
roscopy time compared to case-matched controls. 100% of ques-
tionnaire responses indicated physicians would recommend the AR
system and felt it neither interfered with safety nor increased ra-
diation, contrast, or procedure time. 79% indicated it helped them
navigate through vasculature. 93% indicated it was useful to see
the 3D model in AR.
Conclusions: AR guidance for neuroangiography produced
clinical outcomes, fluoroscopy times, and radiation doses compa-
rable to those of conventional neuroangiography in matched con-
trols. Results suggest that this technology is feasible and safe to use
intraoperatively, offering an opportunity to enhance navigation
through patient anatomy.
4:03 PM Abstract No. 30
Augmented virtual reality assisted treatment
planning for splenic artery aneurysms: a pilot study
Z. Devcic
1
, I. Idakoji
1
, A. Kesselman
1
, R. Shah
1
,
M. AbdelRazek
1
, N. Kothary
1
;
1
Stanford University
Medical Center, Stanford, CA
Purpose: To evaluate the utility of augmented virtual reality (VR)
in preprocedural planning for endovascular repair of splenic artery
aneurysms (SAA) as compared to standard volume-rendering (SR)
software.
Materials: Preprocedural computed tomographic angiography
(CTA) images of 14 patients with 17 SAA who had undergone
endovascular repair were reconstructed using True 3D (EchoPixel,
Inc., CA), a VR visualization software system. AquariusNet (TeraR-
econ, CA) was used for standard volume-rendering image interpreta-
tion. Three radiologists independently evaluated the number of inflow
and outflow arteries using both VR and SR. Procedural angiographic
images served as the gold standard. Improvement in operator confi-
dence of VR over SR was measured on a four-point scale (1 no
change, 4 significant). Clinical utility was objectively measured by
VR’s ability to accurately identify all inflow and outflow arteries
associated with the SAA and subjectively by operator confidence.
Results: There were 17 inflow and 22 outflow arteries associated
with the SAA. The overall sensitivity, accuracy and positive pre-
dictive value for VR was similar to that of SR (91.3%, 89.7%, 84%
and 88.9%, 88.9%, 84.6%, p ¼0.14, respectively). However, the
ability to view and manipulate images in true three-dimensions
using VR markedly improved operator confidence with 93%
receiving a score of at least 3 (71% ¼3, 21% ¼4).
Conclusions: SAA have complex anatomy necessitating metic-
ulous preprocedure planning. VR allows holographic visualization
of images as if they were real physical objects, providing infor-
mation critical for endovascular repair of SAA and thus signifi-
cantly increasing operator confidence.
4:12 PM Abstract No. 31
Efficacy of the preoperative planning for TEVAR
using the greater curvature measurement with
virtual stentgraft image
S. Iwakoshi
1
, S. Ichihashi
2
, S. Sakaguchi
3
,
K. Kichikawa
3
;
1
Nara Medical University, Kashihara
City, Japan;
2
Nara Medical University, Nara, Kashihara,
Japan;
3
Nara Medical University, Kashihara, Nara,
Japan
Purpose: To assess the accuracy of preoperative planning for
TEVAR using the greater curvature measurement with virtual
stentgraft image.
Materials: From January 2012 to December 2016, patients treated
at our institution were retrospectively analyzed. Patients who were
treated with more than two devices, treated for aortic dissection,
and did not have proper preoperative and postoperative CT data
were excluded. From the preoperative CT data, the virtual sten-
tgraft images based on the center lumen line (CL) measurement,
the greater curvature (GC) measurement and the smaller curvature
(SC) measurement were created using SYNAPSE VINCENT
software. These virtual stentgraft images were superimposed on the
postoperative CT to measure the misalignment between these vir-
tual stentgraft images and the actual stentgraft position. A statistical
comparison using Wilcoxon’s signed rank sum test was performed.
In addition, the actual stentgraft lengths were measured based on
CL from postoperative CT data and compared to its original length.
Results: A total of 35 cases were analyzed. Twenty-six were men.
The average age of the patients was 72.4 ±13.0 years. Aneurysms
were located at the descending aorta (n ¼11), and the aortic arch
(n ¼24). The gap between the virtual stentgraft based on SC, CL,
JVIR ▪Scientific Sessions Sunday ▪S17
SUNDAY: Scientific Sessions