Novel Cu/Al2O3-ZrO2 composite for selective hydrogenation of levulinic acid to Γ-valerolactone

Article abstract of DOI:10.1016/j.catcom.2019.03.029

Novel Cu/Al₂O₃-ZrO₂ composite catalyst was developed for the catalytic hydrogenation of levulinic acid (LA) to γ-valerolactone (GVL) in water. By increasing Al₂O₃ content of the Cu/Al₂O₃-ZrO₂ composite up to 5.0 wt% resulted in a nearly 3-fold increase in activity, without loss of selectivity towards GVL. This superior catalytic activity can be attributed to the dilution of Cu by Al₂O₃ as well as the enhanced electronic interaction between Cu and ZrO₂ compared to Cu/ZrO₂ catalyst. Reusability test for four consecutive runs showed a stable activity and selectivity, thus confirming the scope for practical application.

Full text of DOI:10.1016/j.catcom.2019.03.029

Journal of Robotic Surgery
https://doi.org/10.1007/s11701-019-00950-1
ORIGINAL ARTICLE
Eye gaze of endoscopists during simulated colonoscopy
Wenjing He1 · Simon Bryns1 · Karen Kroeker2 · Anup Basu3 · Daniel Birch4 · Bin Zheng1
Received: 4 March 2019 / Accepted: 23 March 2019
© Springer-Verlag London Ltd., part of Springer Nature 2019
Abstract
Regaining orientation during an endoscopic procedure is critical. We investigated how endoscopists maintain orientation
based on video and eye gaze analysis. Novices and experts performed a simulated colonoscopy procedure. Task performance
was assessed by completion time, total distance traveled, maximum depth of insertion, percentage of mucosa viewed, and air
insufation volume. Procedure videos were analyzed by transfers among three viewing areas: center of bowel lumen, edge
of bowel lumen, and other structure without bowel lumen in sight. Performers’ gaze features were also examined over these
viewing areas. Experts required less time to complete the procedure (P<0.001). Novices’ scope traveled a greater distance
(P<0.001) and more scope was inserted compared to an expert (P<0.001). Novices also insufated more air than experts
(P<0.001). Experts maintained the view of bowel lumen in the middle of the screen, while novices often left it on the edge
(P=0.032). When disorientation happened, novices brought the view to the edge more frequently than the center. However,
experts were able to bring it back to the center directly. Eye tracking showed that the rate of saccades in experts increased
when the bowel lumen moved away from the central view, such a behavior was not observed in novices. Maintaining a
centered view of the bowel lumen is a strategy used by expert endoscopists. Video and eye tracking analysis revealed a key
diference in eye gaze behavior when regaining orientation between novice and experienced endoscopists.
Keywords Endoscopy · Eye Movements · Training · Orientation · Competence
Introduction
the scope by coordinating the video images on the monitor
to his/her hand movements. Disorientation or “getting lost”
is a common experience even for expert endoscopists [4].
The term “disorientation in endoscopy” refers to a loss of
both global and local spatial references when navigating the
scope inside the GI tract [5]. When this occurs, the tip of
the scope might apply an inappropriate amount of force to
the intestinal wall, which can lead to undesirable outcomes,
such as bleeding, perforation of the GI tract, or inability to
complete the procedure [4, 6, 7].
Over the last few decades, fexible endoscopy has played a
preeminent role in the diagnosis and treatment of GI diseases
[1]. However, this can be a challenging tool to master, and
physicians in-training must develop a number of fundamen-
tal motor skills to efectively master its use [2, 3]. During
an endoscopic procedure, the performer needs to manipulate
* Bin Zheng
The construct validity of endoscopic VR simulator has
been validated. It has been showed that the performance on
the various tasks on the simulator corresponds to the respec-
tive level of endoscopic experience [8–10]. A shorter proce-
dure time was observed after training for a period of time for
novices [9]. And the skills acquired through VR simulation
transfer well into clinical colonoscopy environment [9, 11].
Currently, there is limited research on endoscopists’ eye
gaze. In the limited studies, eye tracking has been used to
investigate image perception [12] and the eye gaze hot spots
of novice and expert endoscopists [13]. However, there is
no research on gaze behaviors during scope disorientation.
bin.zheng@ualberta.ca
1
Surgical Simulation Research Lab, Department of Surgery,
Faculty of Medicine and Dentistry, University of Alberta,
162 Heritage Medical Research Centre, 8440 112 St. NW,
Edmonton, AB T6G 2E1, Canada
2
2-40 Zeidler Ledcor Centre, Division of Gastroenterology,
Department of Medicine, University of Alberta, Edmonton,
AB, Canada
3
Department of Computing Science, University of Alberta,
Edmonton, AB, Canada
4
Department of Surgery, Centre for the Advancement
of Minimally Invasive Surgery (CAMIS), University
of Alberta, Edmonton, AB, Canada
Vol.:(0123456789)
1 3

Journal of Robotic Surgery
It is still unclear how endoscopists acquire visual input
when navigating the GI track; nor do we understand how
endoscopists use visual cues to re-center the scope when
disorientation occurs.
Participants
Twenty medical students and university students at the Uni-
versity of Alberta were enrolled as novices. Six gastroen-
terologists were recruited as experts for this study. Novices
must have performed less than ten endoscopic procedures or
were without prior endoscopy experience. Experts must have
performed over 200 cases of diagnostic [14] and therapeutic
endoscopy before enrolling in the study.
Our study investigated endoscopists’ eye gaze based on
diferent viewing conditions, which were caused by the
movement of the scope; for example, the moment when the
bowel lumen was in the center of the scene, on the edge of
the scene or was totally out of view. We traced the eye gaze
of novice and expert endoscopists while performing a colo-
noscopy procedure under a virtual reality (VR) endoscopy
simulator. In this study, we would analyze experts’ and nov-
ices’ performance based on video analysis and eye tracking
analysis.
AccuTouch Endoscopy Simulator
The AccuTouch Simulator was renamed as CAE EndoVR
Simulator (Fig. 1a. CAE Healthcare, Montreal, Quebec,
Canada) [15, 16]. The simulator provides users with a real-
istic audio and haptic feedback during an endoscopy pro-
cedure. Modules that include esophagogastroduodenoscopy
(EGD), colonoscopy and endoscopic retrograde cholangio-
pancreatography (ERCP) are available. Endoscopy starts
with patient’s history and changing of various parameters
during the endoscopy, such as vital signs and patient’s pain
response.
We hypothesized that experts will complete the procedure
in a shorter period of time than novices; experts would have
less chance of losing orientation and would perform more
active visual search at the moment of losing orientation than
novices.
Methods
A controlled laboratory study was conducted in the Surgical
Simulation Research Lab at the University of Alberta. Meth-
ods used in the experiment were approved by the Health
Research Ethical Board of the University of Alberta. Con-
sent was obtained from each participant before they enrolled
in the study.
Tasks
The simulator provides subjects with a choice of 20 colonos-
copy cases. We selected “Case 1” for the tasks, as this case
provides the participants with the easiest bowel to manage.
Fig. 1 a Experimental apparatus; b snapshots to show three types of viewing patterns (BLC bowel lumen was in the central vision, BLV bowel
lumen was visible but not in the center, BLL bowel lumen was totally lost)
1 3

Journal of Robotic Surgery
This case was with two polyps in the colon; however, pol-
ypectomy was not required for the procedure.
categorized into three distinct viewing patterns, which
included the period of time when the bowel lumen was
seen in the screen center (BLC), bowel lumen was moved
to the edge of the screen but was still visible (BLV), and
bowel lumen was lost (non-visualized) from the screen view
(BLL). Figure 1b delineates these three diferent viewing
categories. To examine the ability of each endoscopist to
maintain scope orientation, each procedure video was ana-
lyzed by the proportion of time in each category and the
number of transitions between each category.
Overall endoscopy performance
Overall endoscopy performance of individuals was assessed
by examining the output from the simulator (Fig. 2, Part 1).
Data collected from the simulation output were: task com-
pletion time (s), depth insertion of the scope (cm), percent-
age of mucosa viewed (%), total path of the scope traveled
(cm) and total air insufation (cc).
Defning BLC and BLV
Eye tracking and scope video recording
As shown in Fig. 1b, the view from the computer screen is
measured as a 14 cm×14 cm square. We arbitrarily defned
the central view as the area in the middle of the total view
based on a typical view of the colon during the simulation.
This could be represented by an inner square measured by
12 cm × 12 cm, which is about 73% of the total viewing
area. In practice, we drew this inner square on a transparent
plastic sheet, and then attached the sheet on the computer
screen to set up the boundary between the central and the
edge portion of the view. During frame-by-frame video
analysis, an instance of BLC was recorded when the lumen
was completely encompassed the inner square; if the bowel
lumen fell outside of the inner square area, an instance of
BLV was recorded.
Eye gaze was tracked by Tobii X2-60 (Tobii Technology
Danderyd, Sweden). Tobii Studio 3.3.2 (Copyright Tobii
Technology, 2017) software was installed in the computer
that drives the AccuTouch Simulator. For each individual,
the scope movement along with the colon was recorded
by Tobii Studio. The videos were used for determining
instances of scope disorientation (Fig. 2, Part 2). Tobii stu-
dio also can export eye gaze metrics for examining fxation
and saccade (Fig. 2, Part 3).
Scope video analysis
Task time was defned as the period of time between the
frst frame when bowel structure appeared and the frame
when the second polyp disappeared. This period of time was
utilized to analyze the changes of spatial orientation of the
subjects.
Annotation of the videos
Using Tobii Studio, we annotated the frst frame when the
bowel tissue appears and defned the type of the view. We
then waited for another type of view to appear, at which
point we annotated the change. For instance, if the frst
frame was BLL, and the second type of view was BLC,
then the time gap between the frst frame of BLL and the
frst frame of BLC would be the time length of BLL. We
Scope videos were reviewed frame-by-frame by the
experimenter to determine the movement of scope diso-
rientation. Disorientation was defined as a scope posi-
tioned against the colonic wall, leaving the lumen of the
bowel at the edge of the view, or non-visualization of the
bowel lumen. Each video was subsequently annotated and
Fig. 2 A fowchart displaying
system connection and date
output (VR virtual reality)
1 3

Journal of Robotic Surgery
did not stop annotating the events displayed in the video
until the fnal frame, at which point the second polyp
disappeared.
(%), total path the scope traveled (cm) and total air insuf-
fation (cc) between novices and experts.
Data from video analysis
Eye gaze analysis
Independent-sample t test was conducted to compare the
frequency and the length percentage of BLC, BLV and BLL
between experts and novices.
Frequency of saccades, saccadic amplitude, frequency of
fxations, and duration of fxations were retrieved from the
eye tracking data. Through the raw eye tracking data, we
calculated the Rate of Saccades by summarizing frequency
of saccades over each viewing condition, and then divided
by the sum of time on each viewing condition. The Rate of
Fixations was calculated in the same way.
Data from eye gaze analysis
A mixed 2 (group: novice vs. expert)×3 (viewing patterns:
BLC, BLV, BLL) ANOVA was performed to test the difer-
ence of eye metrics (Rate of Saccades, Saccadic Amplitude,
Rate of Fixations, Duration of Fixations) over two groups
under three diferent viewing patterns.
Statistical analysis
Three classes of data were collected and analyzed by dif-
ferent statistical models: (1) the performance output from
the VR simulator of experts and novices; (2) video analy-
sis of the experts’ and novices’ performance in regard to
diferent viewing patterns; (3) eye tracking data analysis in
diferent viewing patterns of experts and novices. Statisti-
cal analysis was performed using SPSS 16.0 (SPSS Inc.,
Chicago, IL, USA). Means and standard errors/deviations
were reported for signifcance, with an a priori a level of
0.05.
Results
Overall task performance
Experts completed the entire task significantly quicker
(96.5 15.2 s vs. 1096.6 135.5 s, P < 0.001). The endo-
scope in hands of novices traveled significantly longer
(431.3 50.3 cm vs. 191.0 14.7 cm, P<0.001) and deeper
into the colon (76.7 0.5 cm vs. 95.9 1.3 cm, P=0.012)
than experts. Also, novices insufated more air than experts
(2371 365.5 cc vs. 694.3 102.6 cc, P<0.001) (Table 1a).
Performance output from the VR simulator
Scope navigation
Independent-sample t test was used to compare the signif-
cance of total time in completing the procedure (s), depth
insertion of the scope (cm), percentage of mucosa viewed
Statistically, experts were faster in reaching the polyp than
novices (47.8 9.0 s vs. 387.8 53.6 s, P=0.005). Experts
had higher frequency of BLC (n=2 vs. n=14, P=0.005),
BLV (n=2 vs. n=9, P<0.001) and BLL (n=2 vs. n=11,
Table 1 Overall task performance and three types of viewing patterns
Groups
Total time (s: mean SE)
Depth of insertion
(cm: mean SE)
Mucosa viewed (%)
Total path (cm:
mean SE)
Total air (cc: mean SE)
Expert
Novice
P
96.5 135.5
1096.6 15.2
0.001
95.85 1.3
76.67 0.5
0.012
11.71 0.5
15.46 2.2
< 0.001
431.3 50.3
191.0 14.7
0.015
2371.6 364.5
694.3 102.6
0.040
Groups
BLC
BLV
BLL
Frequency
Length percentage Frequency
Length percentage Frequency
Length
percent-
age
Expert
Novice
P value
14
2
<0.001
70%
51%
0.108
9
2
16%
24%
0.268
11
2
0.001
14%
23%
0.219
< 0.001
BLC bowel lumen was in the central vision, BLV bowel lumen was visible but not in the center, BLL bowel lumen was totally lost
1 3

Journal of Robotic Surgery
Fig. 3 a Average view transitions of novices; b average view transitions of experts (BLC bowel lumen was in the central vision, BLV bowel
lumen was visible but not in the center, BLL bowel lumen was totally lost)
Fig. 4 Correlation between duration of BLC and procedure time
(BLC bowel lumen was in the central vision)
Fig. 5 Interaction between group and viewing patterns in Rate of
Saccades
P=0.001) than novices. Experts kept the view in the center
longer than novices in length percentage (BLC = 70% vs.
51%, P = 0.108); novices had higher proportion of BLV
(24% vs. 16%, P=0.268) and BLL (23% vs. 14%, P=0.219)
(Table 1b).
Eye gaze characteristics
Signifcance was found in Rate of Fixations (P=0.005). Spe-
cifcally, experts have higher Rate of Fixations (2.68 0.58
count/s) than novices (1.87 0.48 count/s).
The dynamic interactions among three viewing patterns
are demonstrated in Fig. 3. When BLV occurred, experts
returned to BLC with higher chance (n = 7, 29%) than
novices (n = 125, 18%, P < 0.001), yet novices had more
chances of entering into BLL from BLV (n=61, 9%) than
experts (n = 0, 0%). When BLL happened, novices had a
similar chance of entering into BLC (n = 158, 24%) and
BLV (n = 59, 9%) as experts (n = 16, 25%; n = 2, 8%). In
addition, the higher the percentage of BLC, the faster the
task was performed. Percentage of BLC to the total task
time dropped in cases where longer performance time was
recorded (r=−0.399, P=0.044) (Fig. 4).
Under three viewing patterns, signifcant efects were
shown in Saccadic Amplitude which was measured in angu-
lar degree (°) (P=0.05) and Rate of Saccades (P<0.001).
Rate of Saccades was higher in BLV (4.41 1.54
count/s) than both in BLL (4.19 1.51 count/s) and BLC
(4.09 1.49 count/s). Saccadic Amplitude was larger in
BLL (2.99° 1.16°) than both BLV (2.55° 0.68°) and
BLC (2.28° 0.79°). Rate of Fixations was higher in
BLV (2.03 0.64 count/s) than both BLC (1.98 0.46
count/s) and BLL (2.02 0.72 count/s). Duration of Fixa-
tions decreased from BLC (349.40 140.32 ms), BLV
1 3

Journal of Robotic Surgery
Table 2 Eye metrics under three viewing patterns between experts and novices
Variables/(mean SD)
Viewing patterns
BLC
P
BLV
BLL
Novice
Expert
4.73 1.31
Novice
Expert
5.81 1.57
Novice
Expert
Rate of saccades (count/s)
Saccadic amplitude (°)
Rate of fxations (count/s)
4.08 1.49
2.26 0.85
1.82 0.47
4.20 1.41
2.51 0.72
1.83 0.58
4.47 2.20
2.86 1.31
1.87 0.65
4.48 1.47 < 0.001
3.56 1.42
2.58 0.61
2.42 1.10
2.58 0.35
2.75 0.28
2.83 0.34
0.330
0.302
0.338
Duration of fxations (ms) 342.71 154.97 298.15 88.31 331.63 154.86 252.48 97.75 323.83 145.50 257.71 46.09
P values reported here are interaction efects between two diferent groups of performers and three viewing patterns
BLC bowel lumen was in the central vision, BLV bowel lumen was visible but not in the center, BLL bowel lumen was totally lost
(326.32 146.31 ms) to BLL (315.06 134.25 ms) but was
not s ignifcant (Fig. 5).
central view. Our results show that the higher percentage
of BLC is associated with a shorter performance time.
Novices had a higher chance of disorientation, which was
validated by longer time in BLL and BLV. Our results
demonstrate that when BLV occurred, experts brought the
scope view back to BLC but novices were more likely to
completely lose view of the lumen (BLL). This suggests
that the capability of transferring from BLV to BLC might
be a behavior marker to diferentiate between experts and
novices. A training protocol focused on regaining orienta-
tion from BLV to BLC in the future can produce a pro-
found efect on the skill acquisition of novices.
Interaction efects between performer groups and the
viewing patterns are displayed in Table 2. Signifcant inter-
action was only found in the Rate of Saccades (P<0.001).
As shown in Table 2, experts’ Rate of Saccades is 5.81 1.57
count/s when the center of lumen was at the edge of the
screen, which was higher than when the lumen was main-
tained in the central view(BLC: 4.73 1.31 count/s) or lost
to view (BLL: 4.48 1.47 count/s). It seemed that experts
performed more active visual search when they noticed the
lumen moved away from the central view than novices. In
novices, the Rate of Saccades slightly increased but not sig-
nifcantly when the lumen was moving away from the central
view (BLC: 4.08 1.49 count/s; BLV: 4.20 1.41 count/s;
and BLL: 4.47 2.20 count/s).
Eye tracking analysis revealed that the rate of saccades
and fxations increased signifcantly in experts when the
bowel lumen was not in the central view. Experts also dis-
played larger saccadic amplitude than novices, which also
means that experts have higher saccadic velocities [17].
This phenomenon becomes more prominent when the bowel
lumen was deviated from the central view. The above evi-
dence indicates that experts performed active visual search
when they noticed the lumen moved away from the cen-
tral view. However, such behavior was not demonstrated
by novices. According to a previous study regarding eye
gaze patterns between novices and experts, novices tend
to look at more about the tool in hand, while experts are
more target oriented [18]. Presumably, during an endoscopy
practice, when bowel lumen moved away from the central
view, experts performed active visual search about the bowel
lumen to get the view back to center; yet novices might be
looking at their tools to adjust tool orientation. Another note-
worthy fnding is that experts had shorter duration of fxa-
tions in any situations than novices, which can be explained
as experts requiring a shorter duration of time in retrieving
information from the video images than novices, and vice
versa. Moreover, when the bowel lumen was presented in the
central view, subjects displayed longer fxations. It seems
that they can concentrate on the center without the need to
frequently scan peripheral sites.
Discussion
Video analysis and eye gaze data presented here provide
objective data to support our hypotheses. It is not surpris-
ing to fnd that novices required more time to complete the
task. The scope in the hands of novices traveled a longer
distance; in other words, novices moved the scope back
and forth more frequently while navigating. Moreover,
novices tended to insert more scope into the colon. The
unnecessary manipulation of the scope did not increase
the percentage of mucosa viewed by the novices, yet it
might cause discomfort for the patient, since it stretches
more of the colonic structure. Another factor that might
cause patient discomfort is air insufation; in this study,
we found novices insufate more air than experts.
Video analysis of performers’ viewing patterns during
colonoscopy revealed that keeping the bowel lumen in the
center is an important technique for the success of per-
forming a colonoscopy. Expert endoscopists are capable of
adjusting the scope based on visual cues from the bowel,
so they are more capable of maintaining the lumen in the
1 3

Journal of Robotic Surgery
2. Ekkelenkamp VE, Koch AD, de Man RA, Kuipers EJ (2016)
Training and competence assessment in GI endoscopy: a system-
atic review. Gut 65:607–615
Teaching endoscopy has been modeled mathematically
in literature showing that the rate of the improvement of a
trainee’s endoscopic skill is dependent on both the fellow’s
and attendant’s cognitive and endoscopic skill [19]. In focus
groups of gastroenterology fellows about endoscopic teach-
ing, one of the themes identifed was that attending gas-
troenterologists would beneft from instruction on teaching
[20]. Through this study, the future training of endoscopists
should be more specific. For example, the training of
endoscopists could be focused on creating scenes where the
bowel lumen is not well displayed, and then train the novice
endoscopist to bring the view to the central view.
3. Allen JI (2012) Quality assurance for gastrointestinal endoscopy.
Curr Opin Gastroenterol 28:442–450
4. Moore LE (2003) The advantages and disadvantages of endos-
copy. Clin Tech Small Anim Pract 18:250–253
5. Cao CGL, Milgram P (2000) Disorientation in minimal access
surgery: a case study. HFES Ann Meet Proc 44:169–172
6. Luning TH, Keemers-Gels ME, Barendregt WB, Tan AC, Rosman
C (2007) Colonoscopic perforations: a review of 30,366 patients.
Surg Endosc 21:994–997
7. Rabeneck L, Paszat LF, Hilsden RJ et al (2008) Bleeding and
perforation after outpatient colonoscopy and their risk factors in
usual clinical practice. Gastroenterology 135:1899–1906 (e1891)
8. van Dongen KW, Tournoij E, van der Zee DC, Schijven MP,
Broeders IAMJ (2007) Construct validity of the LapSim: can the
LapSim virtual reality simulator distinguish between novices and
experts? Surg Endosc 21:1413–1417
A conscious awareness of the trainee’s eye gaze patterns
represents a cognitive skill that could be used to help fel-
lows improve their endoscopic skill more quickly. To our
knowledge, focusing on where the fellow is looking on the
screen has not been described previously in the literature as
a training tool for teaching endoscopy. However, the difer-
ences in viewing patterns between the novices and experts in
our study provide the basis for testing this theory in a future
study. More works need to be done to relate eye gaze pattern
with expertise level. Another possible research direction is
to use the recordings of the performance to separate various
expertise levels.
9. Triantafyllou K, Lazaridis LD, Dimitriadis GD (2014) Virtual
reality simulators for gastrointestinal endoscopy training. World
J Gastrointest Endosc 6:6–12
10. Seymour NE, Gallagher Anthony G, Roman Sanziana A et al
(2002) Virtual Reality training improves operating room perfor-
mance: results of a randomized, double-blinded Study. Ann Surg
236:458–464
11. Ahlberg G, Hultcrantz R, Jaramillo E, Lindblom A, Arvidsson
D (2005) Virtual reality colonoscopy simulation: a compulsory
practice for the future colonoscopist? Endoscopy 37:1198–1204
12. Meining A, Atasoy S, Chung A, Navab N, Yang GZ (2010) “Eye-
tracking” for assessment of image perception in gastrointestinal
endoscopy with narrow-band imaging compared with white-light
endoscopy. Endoscopy 42:652–655
There are also some limitations within this study: a lack
of an intermediate group of endoscopists to show the con-
tinuous improvement of task performance and eye motions,
which can also be a future direction of research. Another
possible limitation is that the sample size is relatively small,
as a fnite expert’s population are available for this research.
In conclusion, this is the frst step for us to analyze the
performers’ behavior in maintaining the scope orientation
in colonoscopy. Knowledge gained will help us to design a
training protocol for novices to navigate the endoscopy more
efciently, and create an objective assessment of exper-
tise depending on their ability in maintaining the scope’s
orientation.
13. Dik VK, Hooge IT, van Oijen MG, Siersema PD (2016) Measur-
ing gaze patterns during colonoscopy: a useful tool to evaluate
colon inspection? Eur J Gastroenterol Hepatol 28:1400–1406
14. Habaz I et al. (2019) Adaptation of the fundamentals of lapa-
roscopic surgery box for endoscopic simulation: performance
evaluation of the frst 100 participants. Surg Endosc. https://doi.
org/10.1007/s00464-018-06617-6
15. Bar-Meir S (2000) A new endoscopic simulator. Endoscopy
32:898–900
16. Dunkin B, Adrales GL, Apelgren K, Mellinger JD (2007) Surgical
simulation: a current review. Surg Endosc 21:357–366
17. Britannica E (1987) Sensory reception: human vision: structure
and function of the human eye. Google Scholar, Encyclopedia
Britannica
18. Benjamin L, Atkins MS, Kirkpatrick AE, Alan JL (2004) Eye
gaze patterns diferentiate novice and experts in a virtual laparo-
scopic surgery training environment. In: Proceedings of the 2004
symposium on Eye tracking research and applications, ACM, San
Antonio, Texas, pp 41–48
Acknowledgements We thank all the students and gastroenterologists
for participating in this study.
Compliance with ethical standards
19. Sonnenberg A (2018) Limitations of teaching endoscopy. Eur J
Gastroenterol Hepatol 30:252–256
Conflict of interest Dr. Wenjing He, Simon Bryns, Karen Kroeker, Bin
Zheng have no conficts of interest or fnancial ties to disclose.
20. Zanchetti DJ, Schueler SA, Jacobson BC, Lowe RC (2016) Efec-
tive teaching of endoscopy: a qualitative study of the perceptions
of gastroenterology fellows and attending gastroenterologists.
Gastroenterol Rep 4:125–130
References
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional afliations.
1. Varadarajulu S, Banerjee S, Barth BA, Desilets DJ, Kaul V, Kethu
SR, Pedrosa MC, Pfau PR, Tokar JL, Wang A, Wong Kee Song
LM, Rodriguez SA (2011) GI endoscopes. Gastrointest Endosc
74(1–6):e6
1 3