Effects of Artificial Force Feedback in Laparoscopic Surgery
Training Simulators
Audrey K. Bell, Caroline G. L. Cao, Member, IEEE
Abstract— The use of haptic devices to provide artificial
force feedback in teleoperation has been shown to enhance
performance. An experiment was conducted to examine how
artificial (simulated) force feedback is utilized, compared with
real force feedback, in a laparoscopic tissue-probing task.
Actual (real) forces in probing a double-layer silicon gel mass
were replicated and exaggerated in a virtual environment using
a haptic device. Ten subjects performed the probing task in
four different artificial force feedback conditions: 1) high
fidelity force feedback, 2) proportionately exaggerated
feedback, 3) disproportionately exaggerated relative force
feedback, and 4) reversed disproportionately exaggerated
relative force feedback. Results showed that a higher maximum
force was applied, detection time was longer, and distance error
was greater in virtual probing compared to real probing.
Detection time was significantly greater in the virtual high
fidelity condition compared to the disproportionately
exaggerated force feedback conditions. These results suggest
that artificial force information may be processed differently
than real haptic information, leading to higher force
application that could potentially damage tissues, lower
efficiency, and reduced accuracy in tissue probing tasks.
L
I. INTRODUCTION
aparoscopic surgery is a technique that promises less
trauma, reduced scarring, and shorter hospitalization
time compared to open abdominal surgery [1]. Despite
its significant advantages for the patient, laparoscopic
surgery has limitations for the surgeon, such as lack of depth
perception and force feedback, which can make it more
difficult to perform surgery [2]. For this reason, a surgeon
must undergo a considerable amount of training in order to
acquire the unique skills necessary for a successful
laparoscopic procedure [3].
Traditionally, surgeons have relied on the method of
“learning by doing” in the operating room to develop skills.
However, quality control and patient safety issues are
causing this practice to become increasingly less acceptable,
with surgical simulators emerging as a more favorable
training option [3,4]. Virtual reality (VR) simulation is
expected to be a key technology for use in surgical
simulation training [5]. Physical training boxes with typical
tasks such as picking up a pea, while aiding a surgeon in
becoming accustomed to important motor skills, often lack
the visual context of anatomical characteristics that can be
achieved through VR graphics [6]. Additionally, VR
simulators require little maintenance and provide users with
Manuscript received March 1, 2007. This work was supported in part by
the National Science Foundation Grant 0238284.
A.K. Bell and C.G.L. Cao are with Tufts University Department of
Mechanical Engineering, Medford, MA 02144 USA (phone: 617-627-2484;
fax: 617-627-3058; e-mail: [email protected], [email protected].)
1-4244-0991-8/07/$25.00/©2007 IEEE
immediate and objective feedback [3]. In addition to
realistic graphics, an essential feature of effective virtual
reality simulation is high fidelity or realistic haptic feedback.
Since surgeons rely on haptic feedback, or touch sensation,
during delicate surgical procedures, simulators with haptic
feedback are expected to offer more realistic and useful
training than those providing graphics alone [5,7].
As VR simulators are still in their infancy, several options
are being explored for the addition of haptic feedback [6,8].
One way to provide haptic feedback is through a
force-feedback device. These haptic devices have been
incorporated in several surgical simulators and evaluated in
simulated medical procedures, including hernia repair,
epidural completion, and tumor detection [9,10,11].
However, few of these simulators have been assessed to
determine whether the forces created with the haptic device
are processed by the user in the same way as the physical
forces they are meant to simulate.
Previous research has investigated the effect of direct
force feedback, pseudo-force feedback, and redundant force
feedback on grasping force regulation [12]. The results of
these studies showed that pseudo feedback, that is feedback
given through visual or auditory cues, was not as effective as
direct haptic feedback for relaying force information.
Redundant force feedback, or force feedback augmented by
auditory cues, has been shown to be effective for controlling
tissue-probing forces [12, 13]. However, it has been shown
that artificial or simulated force feedback (i.e., force
feedback from a haptic device) can be distorted by the
device’s dynamics [14]. The instability that can arise from
the human grasp has also been identified as a factor that can
affect the efficacy of haptic systems. For example, the
human grasp can absorb mechanical energy, thereby
stabilizing an otherwise unstable system. Alternatively, the
human grasp can reflect energy back into a system that
would otherwise be stable, causing it to become unstable
[15]. Some researchers have tried to design couplings that
will eliminate the system destabilization by using selective
combinations of virtual environment and passive human
operator characteristics [16].
Our primary objective was to determine whether the user
perceives the artificial haptic information and interacts with
it in the same manner as with real-life haptic feedback in a
laparoscopic tissue-probing task. A secondary objective was
to determine whether absolute force magnitude or relative
force gradient was more useful for differentiating between
tissues of varying compliance. That is, our goal was to
determine whether tissue differentiation is made easier by
exaggerating the amount of force feedback in both tissue
layers by the same amount, or by increasing the difference in
2239
the force feedback values between the two layers of tissue.
The performance measures of interest were maximum force
application, time to detection, and the ability to differentiate
between tissue masses of differing compliance, or softness.
Based on previous research, it was hypothesized that the
absolute magnitude of force would be more useful than force
gradient (the difference in hardness between two different
layers of tissue) for tissue compliance differentiation [13].
whose position could be manipulated using the laparoscopic
tool (Fig. 3).
II. MATERIALS AND METHODS
A. Subjects
Ten subjects (5 males and 5 females) having little to no
prior experience using with laparoscopic surgical simulators
participated in the study. The subjects, aged 20-40, were
either right-handed or ambidextrous.
B. Apparatus
The experimental setup consisted of a Phantom Premium
1.0 force-feedback device, a desktop computer displaying
visual graphics, and a laparoscopic grasper. The Phantom
device was attached to the laparoscopic grasper, which
controlled the movement of a needle-shaped stylus in the VR
environment (Fig. 1).
Fig. 2. Experimental setup of Schoonmaker & Cao [13].
C. Procedure and Task
Following the procedure reported by Schoonmaker and
Cao [13], each subject completed a training session the day
before data collection took place. During this training
session, subjects completed three trials of the probing task
under each of the four experimental compliance conditions.
This allowed subjects to become familiar with the simulator,
and helped to avoid discrepancies in the data due to
differences in learning styles. During testing, subjects used
the laparoscopic tool to position the needle and push through
the entry point in the tissue. They were asked to identify the
location at which the top layer of tissue had been surpassed
and the bottom layer of tissue had been reached, based on
when they felt a changing amount of resistance with the
grasper.
Fig. 1. Experimental setup.
The VR environment was a replication of the real
environment used in Schoonmaker and Cao’s study [13].
That real environment presented novice subjects with a dish
of simulated tissue containing two layers of silicone gel of
differing compliance. Subjects were asked to probe the
tissue layers with a laparoscopic needle and locate the
transition point between the two gel layers based on the
force feedback (Fig. 2).
The virtual reality environment was created using
Handshake proSENSETM software. It consisted of four
impenetrable surfaces, simulating walls and a counter top,
which supported a circular object, simulating the disc of
tissue. As with the real tissue, the virtual tissue mass was
characterized by two layers of differing compliance. The
surface of the virtual tissue was marked with a target entry
point. The virtual environment also contained a needle
Fig. 3. Virtual task scene.
D. Experimental Design
A 4 Compliance x 5 Tissue-thickness, within-subjects
repeated measures design was used. The four compliance
conditions were: 1) high fidelity condition, simulating a
physical double-layer gel mass whose two layers had
compliance similar to human liver and fatty tissue,
respectively; 2) proportionately exaggerated condition,
where the amount of friction in the two tissue layers was
increased, but the relative friction between the two layers
was unchanged; 3) disproportionately exaggerated condition,
where both the intensity of the friction and the change in
2240
friction between the two tissue layers were increased; and 4)
reversed disproportionately exaggerated condition, where
the friction coefficients in the gel layers were the same as
those in the disproportionate exaggerated condition, but the
top and bottom layers were inverted. In this condition, the
upper layer of gel was less compliant than the bottom layer,
whereas every other condition placed the softer layer on top.
The purpose of the high fidelity condition was to create a
simulation that was the virtual equivalent to the setup used
by Cao and Schoonmaker [13].
The proportionately
exaggerated and disproportionately exaggerated conditions
were included so that the effect of varied force magnitude
could be compared with the effect of varied force gradient
(the difference in the amount of friction between the two
layers).
The reversed disproportionately exaggerated
condition was created to investigate whether the order in
which compliance layers are encountered plays a role the
detection of the compliance change.
Five different tissue thicknesses were used to avoid a
learning effect. Subjects performed three trials with each
tissue thickness in each of the four conditions for a total of
60 trials. The order of tissue thickness and conditions were
completely randomized.
E. Dependent Measures
The dependent measures in the experiment were time to
detection, maximum force application, and error in
detection. Error was measured by the deviation from the
interface between the two tissue layers.
F. Data Analysis
Data were analyzed using analysis of variance (ANOVA)
and t-tests. In addition, performance on the high fidelity
force-feedback condition was compared to results in the real
environment from Schoonmaker & Cao [13].
was also a significant difference shown between the regular
and reversed conditions with exaggerated relative friction,
where less force was used in the reversed condition
(p=0.002).
Fig. 4. Penetration depth error. *Data obtained from Schoonmaker & Cao
[13].
C. Detection Time
T-tests revealed that detection time, defined as the time
elapsed from the beginning of the task to the moment at
which the subject located the second layer of gel, was
similar in the high fidelity and exaggerated conditions, but
was significantly lower in the disproportionately
exaggerated and the reversed disproportionately exaggerated
conditions (p<0.001) (Fig. 6). In addition, an ANOVA test
showed that subjects completed the real tasks in a
significantly shorter amount of time than the high fidelity
tasks (p<0.001).
IV. RESULTS
A. Penetration Depth Error
When performance error with artificial feedback was
compared to performance error with real haptic feedback, an
ANOVA test revealed that subjects were significantly more
accurate in locating the tissue layer transition point when
probing the real gel (p=0.007) (Fig. 4). In the artificial force
conditions, error in detection varied with the intensity of the
feedback. T-tests showed that error of detection was
significantly lower in both disproportionately exaggerated
force conditions when compared with the high fidelity
condition (p<0.001, p=0.001), and with the proportionately
exaggerated condition (p=0.002, p=0.007).
B. Force Application
An ANOVA test showed that the average maximum force
applied in the high fidelity virtual environment was
significantly higher than that in the real environment
(p<0.001) (Fig. 5). In addition, t-tests showed that
significantly more force was used in the proportionately
exaggerated
condition
than
in
the
reversed
disproportionately exaggerated condition (p<0.001). There
Fig. 5. Maximum force application. *Data obtained from Schoonmaker &
Cao [13].
Fig. 6. Average detection time. *Data obtained from Schoonmaker & Cao
[13].
2241
V. DISCUSSION
One objective of this study was to determine whether
artificial force feedback is similar to real haptic feedback for
task performance.
Our results show that subjects
consistently used a larger amount of force, a larger amount
of time, and were significantly less accurate in the virtual
environment than the real environment, even though the
forces in both environments were equal in magnitude. This
suggests that real and artificial force information may be
processed differently. Alternatively, the difference may be
due to variations in the two environments. There may be
other interaction forces from direct contact with real gels
that are missing from the haptic device. It is also possible
that the grasp that the subjects had on the tool caused the
information to be transferred between the device and the
subject differently than it would have been in a real system
[15]. Therefore, the perception-action loop in compliance
differentiation may be more complex than previously
thought. That is, other cues such as tactile feedback may
also play a role. Subjects were given only force feedback
through the haptic device, which lacks other redundant
sensory information such as vibrotactile or auditory
feedback. These results indicate that haptic feedback
devices such as the Phantom should be tuned to more
accurately represent the type of force-feedback experienced
in real life before they can be utilized as accurate surgical
training devices.
A second objective of this study was to determine
whether force magnitude or force gradient was more useful
for differentiating between tissues of varying compliance.
Previous research suggests that when using vibrotactile
feedback to enhance force perception, subjects rely more on
the magnitude of vibration than intensity changes [13].
Based on these findings, we expected that the compliance of
the tissue would be a more important cue than the difference
in compliance between the two layers of tissue; that is, we
expected that subjects would perform better in tasks with
proportional exaggeration than those with disproportional
exaggeration. In fact, the opposite was found. Subjects
were able to detect the difference in compliance significantly
more rapidly and with significantly greater accuracy when
faced with a larger force gradient. This indicates that
subjects were aided more by a large difference in
compliance between the two tissues than they were by an
increase in force feedback alone.
VI. CONCLUSION
The results of the study indicated that force gradient is
more important than force magnitude for tissue
differentiation. That is, it is easier to differentiate between
tissues that have a large difference in compliance than
tissues that are both relative hard. This information may
allow for the design of more effective training exercises for
laparoscopic surgeons. With respect to the nature of force
feedback, it was observed that participants consistently used
more force, were less efficient, and had reduced accuracy
when completing a laparoscopic task in a virtual
environment with artificial force feedback, compared to the
same laparoscopic task performed in a real environment.
These results suggest that artificial force information may be
processed differently from real force feedback. Therefore,
artificial force feedback mechanisms must be more precisely
tuned in order to be used as equivalents in laparoscopic
surgical training exercises. The transfer of training from
surgical simulators with unmatched artificial force feedback
to real tissue manipulation in the OR is yet undetermined.
REFERENCES
[1] S. Maithel, R. Sierra, J. Korndorffer, P. Neumann, S.
Dawson, M. Callery, et al. “Construct and face validity
of MIST-VR, Endotower, and CELTS,”
Surgical
Endoscopy, vol. 20, no. 1, pp. 104-112, Jan. 2006.
[2] A. C. W. Chan, S.C.S. Chung, A.P.C. Yim, J.Y.W. Lau,
E.K.W. Ng, A.K.C.Li, “Comparison of two-dimensional
vs three-dimensional camera systems in laparoscopic
surgery,” Surgical Endoscopy, vol. 11, no.5, pp.
448-440, May 1997.
[3] Y. Munz, B.D. Kumar, K. Moorthy, S. Bann, A. Darzi,
“Laparoscopic virtual reality and box trainers: Is one
superior to the other?” Surgical Endoscopy, vol. 18, no.
3, pp. 485-494, Mar. 2004.
[4] J. Vozenilek, S. Huff, M. Reznek, J. Gordon, “See One,
Do One, Teach One: Advanced Technology in Medical
Education,” Academic Emergency Medicine, vol. 11,
no. 11, pp. 1149-1154, 2004.
[5] Y. Kuroda, M. Nakao, T. Kuroda, H. Oyama, M.
Komori, “Interaction model between elastic objects for
haptic feedback considering collisions of soft tissue,”
Computer Methods and Programs in Biomedicine, vol.
80, pp. 216-224, 2005.
[6] V. Vuskovic, M. Kauer, G. Szekely, M. Reidy,
“Realistic Force Feedback for Virtual Reality Based
Diagnostic Surgery,” Robotics and Automation, 2000.
Proceedings. ICRA. IEEE International Conference,
vol. 2, pp. 1592-1598, 2000.
[7] G. Burdea, “Haptics Issues in Virtual Environments,”
Computer Graphics International, 2000. Proceedings,
pp. 295-302, 2000.
[8] O. Reich, M. Noll, C. Gratzke, A. Bachmann, R.
Waidelich, M. Seitz, et al, “High - Level Virtual Reality
Simulator for Endourologic Procedures of Lower
Urinary Tract,” Adult Urology, vol 67, no. 6, pp.
1144-1148, Jun. 2006.
[9] T. Dang, T.M. Annaswamy, M.A. Srinivasan,
“Development and evaluation of an epidural injection
simulator with force feedback for medical training,”
Studies in Health Technology and Informatics, vol. 81,
pp. 97-102, 2001.
[10] R.J. Lapeer, P. Gason, J.L. Florens, S. Laycock, V.
Karri, “Introducing a novel haptic interface for the
planning and simulation of open surgery” Studies in
Health Technology and Informatics, vol. 98, pp. 197-9,
2004.
[11] J. Stalfors, T. Kling-Peterson, M. Rydmark, T. Westin,
“Haptic palpation of head and neck cancer patients –
2242
implication for education and telemedicine,” Studies in
Health Technology and Informatics, vol. 81, pp.
471-474, 2001.
[12] P. Richard, G, Birebent, P. Coiffet, G. Burdea, D.
Gomez, N. Langrana, “Effect of frame rate and force
feedback on virtual object manipulation,” Presence:
Teleoperators and Virtual Environments, vol. 5, no.1,
pp. 95-108, 1996.
[13] R. Schoonmaker, C. Cao, “Vibrotactile force feedback
system for minimally invasive surgical procedures,”
Proceedings of the 2006 IEEE International Conference
on Systems, Man, and Cybernetics, pp. 2464-2469,
2006.
[14] M. Moreyra, B. Hannaford, “A practical measure of
dynamic response of haptic devices,” Proceedings of
the 1998 IEEE International Conference on Robotics &
Automation, vol. 1, pp. 360-374, May 1998.
[15] R. Adams and B. Hannaford, “Stable haptic interaction
with virtual environments,” IEEE Transactions on
Robotics and Automation, vol. 15, no. 3, pp 465-474,
Jun. 1999.
[16] R. Adams and B. Hannaford, “A two-port framework
for the design of unconditionally stable haptic
interfaces,” Proceedings of the IEEE/RSJ International
Conference of Intelligent Robots and Systems, pp.
1254-1259, 1998.
2243