Introduction
Colorectal cancer (CRC) has one of the highest incidences of all cancers in the developed
world [1]. CRC screening and early diagnosis have proven benefit in numerous international
large-scale, population-based studies and secondary care-based retrospective case
control studies. [2]
[3]. Currently, video-colonoscopy is considered the gold standard for diagnosis of colorectal
neoplasia [4]. There are, however, important limitations of this technique, including invasiveness,
patient discomfort and risk of complications [5]
[6].
Trainees entering colonoscopy training programs face a challenging and long skill
acquisition process in order to acquire an appropriate level of dexterity and lesion
recognition. According to some studies, changing to non-conventional modalities integrating
robotic technology with imaging for colonoscopy may influence patient comfort, compliance
as well as health professional and resource demand [7]. New developments within robotic technology include disposable probes that have
potential to be safer and less painful while improving infection control and reducing
costs associated with decontamination services. This may enable early diagnosis of
colorectal pathology in primary care as a result of better engagement with patients.
The robotic colonoscope (RC) is also designed have a more intuitive steering mechanism
than standard colonoscopy, which may facilitate a more approachable diagnostic skill
acquisition process for colonoscopists [8] and enable health professionals to make a diagnosis in the community or primary
care setting.
In this study we assessed the operator acceptability of the new RC with use of a simulation-based
training model. We also investigated the time taken for acquisition of skills by the
participants performing RC.
Materials and methods
Participants
Invitations to participate in the study were sent out to health professionals with
varying degrees of experience and expertise in colonoscopy who were from the hospital
administrative and teaching facilities at Cardiff and Vale University Health Board,
Wales. The participants were split into groups of different knowledge, skill and hand-eye
coordination abilities. Distribution of the participants (n = 9) is shown in [Fig. 1]. We enrolled three expert endoscopists – consultant gastroenterologists, all with
individual experience of a few thousand colonoscopies. The trainees with the intermediate
level of endoscopic knowledge and skills were two specialist endoscopic nurses. The
two novices (with knowledge and related skills but no colonoscopic skills from training
or practice) included a nurse performing upper gastrointestinal endoscopies and a
nurse performing capsule endoscopies. We also recruited two participants with no endoscopy
knowledge or skills but who were adept in video games with joy-stick control.
Fig. 1 Distribution of study participants.
Robotic colonoscope
All procedures were performed with an Endotics disposable probe [9]. The Endotics System is composed of a disposable probe and a workstation ([Fig. 2]). The workstation enables the endoscopist to fully control the disposable probe
with a hand-held console and visualize the lumen on screen with real-time images.
With the hand-held console and via the motor, the colonoscopist is able to steer the
robotic colonoscope in every direction, elongate the body of the probe to move it
forward, and apply rinsing, insufflation, and suction. The option to capture images,
record videos and obtain biopsies is also available. The aim of the initial testing
was to evaluate tip control and precision skills, therefore, more advanced procedures
such as polyp resection were not assessed. That will be addressed in a follow-up study
for which patient recruitment has been completed.
Fig. 2 The robotic colonoscope: the workstation with the probe attached.
Study design
All nine participants performed a colonoscopy with the RC on an adapted “colonoscopy
suitcase” model developed at the Welsh Institute for Minimal Access Therapy (WIMAT)
center. This model has previously been validated and used in training in diagnostic
and therapeutic colonoscopy on numerous courses in the UK over the past 5 years and
provides a realistic luminal appearance [10]. The colon was of the same standard length for all procedures. Twelve polyps were
placed around the colon: one in the rectum, three in the sigmoid, two in the descending
colon, one in the splenic flexure, two in the transverse colon, one in the hepatic
flexure, one in the ascending colon and one in the cecum. All polyps were of similar
size and morphology – flat circular or flat angular lesions.
Before testing, each participant received both verbal and written instructions on
the goals of the study and information about the semi-automated RC along with a familiarization
period with the device on a “training model” ([Fig. 3]). Afterwards participants filled out an in-house-developed questionnaire evaluating
the RC. Of nine invited participants, four participated in a follow-up session to
analyze skill acquisition. The time interval between the initial and follow-up session
was either 6 or 7 days, during which time the participants had no opportunity to practice
the RC.
Fig. 3 Participant performing the procedure. The “training model” used during the familiarization
period is visible on the table in front of the blue screen. The adapted “colonoscopy
suitcase” is placed behind the screen to decrease availability of additional visual
cues for the participant.
Study end points
Study end points were cecal intubation time, withdrawal time and number of detected
lesions on insertion and withdrawal. Participants’ evaluation of the RC was assessed
with a qualitative questionnaire using closed and short answer questions.
Results
Procedure characteristics
On average, experts required a shorter time to intubate the cecum with mean 29 minutes
58 seconds, followed by video gamers (41 min 04 sec), trainees (44 min 28 sec) and
novices (50 min 10 sec), however, there was wide variation within most groups ([Fig. 4]). Video gamers had the shortest withdrawal time with mean 13 minutes 31 seconds,
followed by experts (18 min 26 sec), trainees (19 min 47 sec) and novices (27 min
52 sec) ([Fig. 4]).
Fig. 4 Scatter plots with cecal intubation and withdrawal times for different participant
groups. Each mark demonstrates performance of a single participant. The horizontal
line represents the mean time for each participant group. Circle, expert; triangle,
trainee; square, novice; star, video gamer.
Polyp detection
The overall polyp detection rate (84.26 %) was the highest in the novice group (91.67 %)
followed by the experts (86.11 %) ([Fig. 5]). Both trainee and video gamer groups had a slightly lower polyp detection rate
of 79.17 %. For eight of nine participants, polyp detection was higher during the
scope withdrawal than during scope insertion. The most commonly missed polyps were
placed in the transverse colon (41 % of all missed polyps) and sigmoid colon (35 %).
The rest of the missed polyps were placed in the rectum (12 %) and splenic flexure
(12 %).
Fig. 5 Scatter plot presenting the total polyp detection rate for participants. A rate of
100 % equals detection of 12 simulated polyps. Each mark demonstrates performance
of a single participant. The horizontal line represents the mean detection rate for
each participant group. Circle, expert; triangle, trainee; square, novice; star, video
gamer.
Follow-up session
Four of nine participants attended the follow-up session where they were asked to
repeat the procedure. No additional information about the RC and no assistance with
the joy-stick controller functions was provided during the follow-up session. During
follow-up, all of the participants improved their performance. Each participant had
a lower cecal intubation time during the follow-up session than in the initial session
as well as the same or higher polyp detection rate ([Fig. 6]).
Fig. 6 Comparison of participant performance between initial and follow-up session. One
participant from each category took part in the follow-up session.
Questionnaire evaluation
After completing the procedure, participants filled out a questionnaire evaluating
the RC. [Table 1] shows the participants’ views on future use of RC, with the majority of the participants
perceiving a potential role for RC in an out-of-hospital environment. In terms of
performance, the slow speed of scope advancement due to automatic sequencing was the
most consistently identified drawback of the RC. When that was investigated further,
the majority of participants could see a role for RC in diagnostic procedures ([Table 2]).
Table 1
Subjective evaluation of experience with robotic colonoscope.
|
Statement
|
Total sample
(n = 9),
n (%)
|
Participant expertise
|
|
Expert
(n = 3)
|
Trainee
(n = 2)
|
Novice
(n = 2)
|
Gamer
(n = 2)
|
|
Where do you think the robotic colonoscopy could be best used?
|
|
|
3 (33 %)
|
0
|
1
|
0
|
2
|
|
|
5 (56 %)
|
2
|
1
|
2
|
0
|
|
|
1 (11 %)
|
1
|
0
|
0
|
0
|
|
If this new device was available for clinical use currently and if you were given
appropriate training would you:
|
|
|
4 (45 %)
|
2
|
1
|
0
|
1
|
|
|
4 (45 %)
|
1
|
1
|
2
|
0
|
|
|
1 (10 %)
|
0
|
0
|
0
|
1
|
|
Which of these two options is the greatest drawback for RC?
|
|
|
7 (78 %)
|
2
|
1
|
2
|
2
|
|
|
1 (11 %)
|
0
|
1
|
0
|
0
|
|
|
1 (11 %)
|
1
|
0
|
0
|
0
|
|
Would training on the controller used for the RC be useful for people learning to
use it?
|
|
|
6 (67 %)
|
1
|
1
|
2
|
2
|
|
|
3 (33 %)
|
2
|
1
|
0
|
0
|
Table 2
Evaluation of potential future use of RC.
|
Statement
|
Average rank
|
|
How do you think the robotic colonoscopy could be best used? (rank 1 – 3, one indicates
most preferred)[1]
|
|
|
1.7
|
|
|
1.6
|
|
|
2.1
|
1 One person felt unqualified to rank the statements, and therefore was not included
in this analysis. One person missed the box for training and did not put a rank for
this.
The questionnaire also assessed views regarding RC compared to standard colonoscopy.
Participants were asked to identify and describe in their own words easier/more challenging
and more/less intuitive aspects of RC. The RC received positive feedback for being
less physical to handle and operate from seven participants (77.78 %). It was perceived
that it could be mastered relatively easily, with additional exposure. Specific aspects
of RC that were perceived to be more intuitive compared to standard colonoscopy were
identification of the appropriate direction of movement and setup of the controller.
Slow speed was noted as a consistent feature of RC (55.56 % participants). Intermittent
or occasional suboptimal views were also described as a drawback in comparison to
standard colonoscopy (22.22 % participants). Aspects of RC that were perceived to
be least intuitive compared to standard colonoscopy related to unfamiliarity with
joy-stick controller and lack of manual control over probe position (44.44 %). Less
experienced participants felt comfortable with using RC for any diagnostic procedure.
The experts made suggestions for RC to be used for specific patient groups: cases
in which people experienced too much discomfort from standard colonoscopy, or low-risk
patients.
Discussion
In this study, we assessed operator acceptability and skill acquisition with the new
RC. The results show that all participant groups were able to complete colonoscopy
within the setting of a training simulator, as summarized in [Table 3]. The overall polyp recognition rate by the participants was comparable to a similar
study using a different RC with a joy-stick controller [7]. Cecal intubation time was generally longer than that expected with conventional
colonoscopy, which may reflect the completely different skills involved in the precise
control mechanisms of the RC procedure.
Table 3
Summary of participant outcomes.
|
Parameter
|
Expert 1
|
Expert 2
|
Expert 3
|
Intermediate 1
|
Intermediate 2
|
Novice 1
|
Novice 2
|
Video gamer 1
|
Video gamer 2
|
|
Cecal intubation time? (minutes)
|
20.6
|
44.7
|
24.6
|
26.2
|
75.9
|
62.8
|
24.1
|
57.6
|
24.5
|
|
Withdrawal time? (minutes)
|
15.0
|
19.7
|
20.6
|
17.3
|
22.6
|
22.2
|
33.2
|
15.7
|
11.3
|
|
Total time? (minutes)
|
35.6
|
64.4
|
45.2
|
43.5
|
98.5
|
85.0
|
57.3
|
73.3
|
35.8
|
|
Number of lesions detected during probe insertion?
|
8 (67 %)
|
7 (58 %)
|
7 (58 %)
|
8 (67 %)
|
8 (67 %)
|
6 (50 %)
|
7 (58 %)
|
7 (58 %)
|
8 (67 %)
|
|
Number of lesions detected during probe withdrawal?
|
10 (83 %)
|
10 (83 %)
|
10 (83 %)
|
10 (83 %)
|
9 (75 %)
|
9 (75 %)
|
11 (92 %)
|
6 (50 %)
|
11 (92 %)
|
|
Total number of detected lesions? (%)
|
10 (83 %)
|
11 (92 %)
|
10 (83 %)
|
10 (83 %)
|
11 (92 %)
|
9 (75 %)
|
11 (92 %)
|
8 (67 %)
|
11 (82 %)
|
|
Number of colonoscopies performed?
|
> 5000
|
> 1000
|
> 5000
|
35
|
40
|
0
|
0
|
0
|
0
|
Our pilot data suggest that RC use may be worthy of investigation in a primary or
intermediate care setting. Use of a semi-automated robotic device seems to have potential
for improving diagnostics, expanding training programs and facilitating management
of “difficult cases.” The study suggests that participants using this interface seem
to rapidly improve their performance, regardless of previous colonoscopy experience
in this small cohort.
Intermittent suboptimal views described as a drawback in comparison to standard colonoscopy
may be a feature related to training in and acquisition of skills with the RC. Perceived
lack of control over movement could reflect current clinical practice of using standard
colonoscopy, which enables great levels of control for clinicians with their hands.
It seems this apparent lack of control is viewed negatively. Further challenges may
be related to adapting to the joy-stick controller and the button functions. Console
familiarity and experience of using joystick-controlled mechanisms may need further
standardization at baseline in future follow-up studies. Wide variation in participants’
outcomes may be a result of multiple individual factors and the reasons for it may
become clearer in a larger cohort.
The limitations of the study were its small sample size and relatively limited follow-up
data. The primary aim was to assess operator acceptability with the new device. Ideally,
the sample size would be larger, with more participants in each study group. Another
limitation was lack of direct comparison between RC and standard colonoscopy on the
simulation-based training model and, therefore, lack of ability to comment on RC parameters
such as procedure time or polyp detection. However, as a pilot study to test acceptability
of the RC, this study shows interesting potential of robotics in the field of colonoscopy.
Conclusion
By assessing acceptability of the new device by operators with varying degrees of
knowledge and experience, this project is the first step in assessment of specific
training needs and development of a training program in robotic endoscopy. We are
currently engaged in a pilot evaluation of the feasibility and acceptability of this
technology in a patient cohort, which will assess whether skills acquired on the simulator
will translate to similar outcomes in patients.
