IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO.

12, DECEMBER 2010 2891

Design and Fabrication of a Magnetic Propulsion

System for Self-Propelled Capsule Endoscope
Mingyuan Gao, Student Member, IEEE, Chengzhi Hu, Zhenzhi Chen, Honghai Zhang,
and Sheng Liu∗ , Senior Member, IEEE

Abstract—This paper investigates design, modeling, simulation,

and control issues related to self-propelled endoscopic capsule nav-
igated inside the human body through external magnetic fields.
A novel magnetic propulsion system is proposed and fabricated,
which has great potential of being used in the field of noninvasive
gastrointestinal endoscopy. Magnetic-analysis model is established
and finite-element simulations as well as orthogonal design are
performed for obtaining optimized mechanical and control pa-
rameters for generating appropriate external magnetic field. Sim-
ulated intestinal tract experiments are conducted, demonstrating
controllable movement of the capsule under the developed mag-
netic propulsion system.
Index Terms—Biomedical equipment, capsule endoscope, gas-
trointestinal (GI) endoscopy, magnetic propulsion, orthogonal pro-
cess design.

I. INTRODUCTION
VER 3 million people annually in the U.S. alone suf-
O fer gastrointestinal (GI) disease serious enough to re-
quire hospitalization and 300 000 people die from GI diseases
Fig. 1. Background of capsule endoscopy (GI tract image was from [34]).

annually in China [1]–[3]. Recently, commercialized passive-

controlled wireless capsule endoscopy, which had been devel- scope with the size of a normal pill can be easily swallowed by
oped since 2000, exploits a swallowable capsule endoscope that patients of various ages. It takes pictures throughout the GI tract
includes digital camera, application-specified integrated circuit by sending images to an external recorder, and thus provides
(ASIC) transmitter, antenna, illuminating LED-assisted imag- useful information for clinic diagnosis of GI tract diseases (as
ing, and battery for access to the GI tract [1], [4]. Capsule endo- shown in Fig. 1).
Although capsule endoscopy has achieved great progress
compared to traditional invasive endoscopy, it still has many
limitations [5]–[7]. With the development of electronics, mi-
Manuscript received January 7, 2010; revised April 1, 2010; accepted May crosystems fabrication, and wireless communication technol-
25, 2010. Date of publication June 10, 2010; date of current version November ogy, research and progress on a self-propelled capsule endo-
17, 2010. This work was supported by the National High Technology Research scope have been the topic of interests for several years [8]–[11].
and Development Program of China under the Ministry of Science and Tech-
nology under Program 863 (no. 2008AA04Z313). The work of M. Gao and C. A wide variety of active and wireless control mechanisms have
Hu was supported by the Graduate Innovation Foundation Program from the been developed for self-propelled endoscopic capsule such as
Huazhong University of Science and Technology (no. HF0601108100). M. Gao shape memory alloy (SMA) actuator, spiral-type micromachine,
and C. Hu equally contributing co-first authors. Asterisk indicates correspond-
ing author. electrical stimulus capsule, motor-legged robot, and hydraulic
M. Gao was with the School of Mechanical Science and Engineering, manipulator [12]–[33].
Huazhong University of Science and Technology, Wuhan 430074, China. Ishiyama et al. in Tohoku University proposed one type of
He is now with the Micro Energy System Lab., the University of Tokyo, Japan
(e-mail: [email protected]). micromachine with an inside permanent magnet and a spiral
C. Hu was with the School of Mechanical Science and Engineering, structure [12]–[14]. The micromachine was rotated and pro-
Huazhong University of Science and Technology, Wuhan 430074, China. pelled by applying an external rotational magnetic field, which
He is now with Nagoya University, Nagoya 464-8601, Japan (e-mail:
[email protected]). was generated by three pairs of Helmholtz coils. They had re-
Z. Chen and H. Zhang are with the School of Mechanical Science and En- ported corresponding experiments both in vivo and in vitro,
gineering, Huazhong University of Science and Technology, Wuhan 430074, demonstrating feasibilities of the rotating magnetic guidance
China (e-mail: [email protected]; [email protected]).
∗ S. Liu is with the School of Mechanical Science and Engineering, and the system [15], [16]. However, due to safety issues of electromag-
National Laboratory for Optoelectronics, Huazhong University of Science and netic radiation induced by low-frequency magnetic field and
Technology, Wuhan 430074, China (e-mail: [email protected]). inaccessibility of the spiral-based capsule to the stomach, ap-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. plication of this technique is still limited in minimally invasive
Digital Object Identifier 10.1109/TBME.2010.2051947 therapy of human body.
0018-9294/$26.00 © 2010 IEEE
2892 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 12, DECEMBER 2010

TABLE I
ACTIVE LOCOMOTION APPROACHES OF CAPSULE ENDOSCOPY

Kim et al. developed a locomotion mechanism for endoscopic Preliminary experimental results in vitro demonstrated that the
capsule using SMAs [17]. Prototype endoscopes including four device was capable of traveling in the digestive tract with a
pairs of SMA springs and clampers were fabricated, and in vitro typical speed ranging between 10 and 40 mm/min.
tests confirmed that the device was able to move in the tract Carpi et al. at the University of Pisa developed a solution us-
with a speed of 9.19 mm/min at a current of 300 mA when ing an external magnetic field applied to the endoscopic capsule,
controlling the four pieces of the moving modules sequentially. thus providing an approach of controlling capsule’s location pre-
However, considering issues of great power consumption and viously coated with a magnetic shell [24]–[26]. This technique,
high operation temperature associated with SMA-based micro- on the one hand, can interact with an imposed external magnetic
robot, state-of-the-art power-generation technology is expected field in the GI tract; on the other hand, it is readily and cheaply
to be developed for substituting the currently commercial bat- compatible with currently commercialized capsule endoscope,
tery for this specific application. and thus needs no modification of internal configuration. Re-
Mosse et al. proposed an electrical simulation method for cently, they used a commercial computer-aided surgery system
propelling an endoscope in the small intestine [18]. The loco- (Niobe, Stereotaxis, St. Louis, MO) as the controllable source
motion principle is that, by electro-stimulating circle muscle of magnetic field. They have reported the feasibility of using
contraction of the intestinal wall, the device can be propelled the technology of magnetic navigation in the field of GI en-
forwardly or backwardly. Five devices with a diameter ranging doscopy, and preliminary proof-of-concept investigations have
from 15 to 23 and with a taper of 28–40 degrees were tested in been performed [25]. However, The Stereotaxis is an advanced
the small intestine and esophagus of anesthetized pigs. When cardiology system for use in the cardiac catheterization. Thus,
stimulated at 15 Hz with 30-ms pulses at 20 mA, the device a navigation system should be designed for negotiating the spe-
moved reliably in either direction in the small intestine at a cific application of GI endoscopy. At present, to the best of our
maximum speed of 4.5 mm/s and negotiated tight curves. Elec- knowledge, few automated magnetic navigation systems have
trical stimulus approach has an unknown effect to the peristalsis been proposed and fabricated for serving the patients in the
mechanism of the intestine. Further research needs to be carried field of noninvasive GI endoscopy due to many concerns such
out for inspecting this electrophysiological problem. as safety and cost (Table I is a review of current active locomo-
Dario’s research group proposed an actuation mechanism ex- tion approaches of capsule endoscopy).
ploiting a legged microrobot to perform active exploration of Based on these motivations, we proposed an automated mul-
the digestive tract [19]–[22]. This device is equipped with four tiaxis magnetic propulsion system for specific application of
super elastic legs, allowing large stroke advancement in the GI controlling the endoscopic capsule’s location and orientation
tract, and a CMOS frontal camera to visualize the intestinal wall. in the GI tract. Magnetic-analysis model was established and
GAO et al.: DESIGN AND FABRICATION OF A MAGNETIC PROPULSION SYSTEM FOR SELF-PROPELLED CAPSULE ENDOSCOPE 2893

Fig. 3. Overall assembly drawing of magnetic propulsion system.

the endoscopic capsule less than 5 dollars and dimension of

ø12 mm × 34 mm (without magnetic shell). In this paper, we
mainly focus on discussion of the developed magnetic propul-
sion system as follows.

B. Magnetic Propulsion System

Fig. 3 illustrates the assembly drawing of magnetic propul-
sion system. The patient support, which can be moved along
the longitudinal direction forwardly and rearwardly and along
Fig. 2. Engineering design of the magnetic propelled endoscopic system. (a) the lateral direction inwardly and outwardly, is mounted on a
Illumination module, vision module, controlling module, and wireless transmis- pedestal, allowing the patient to be moved relative to the magnet
sion module are related through flexible connection. The overall dimensions of
the device are 12 mm × 34 mm (without magnetic shell). (b) Schematic diagram assembly. The magnet assembly comprises several Nd–Fe–B
showing operation flow of the magnetic propulsion system. permanent magnets, ball screws, linear ball rails, and gripper
device. The ball screws convert motor’s rotary motion to ball
finite-element (FE) method was developed for obtaining opti- nut’s linear motion, exerting necessary force to move the per-
mized mechanical and control parameters for generating appro- manent magnets along a longitudinal axis. Once the two groups
priate external magnetic field. In order to judge the feasibility of of oppositely positioned magnets are moved in the opposite di-
the proposed system, experimental tests were carried out and it rection parallel to the longitudinal axis, the magnet inside the
demonstrated controlled translations and rotations of the capsule capsule endoscope will be rotated about its lateral or vertical
under this novel propulsion system. axis or any axis in the transverse plane according to the differ-
ent positions of magnet support. The magnet support contains
II. SYSTEM DESIGN AND LOCOMOTION ANALYSIS a pair of worm/worm gear, a pivot, and aluminum framing.
The magnet assembly is mounted onto the magnet support and
In this study, we focus on the design and integration of a
can rotate about the longitudinal axis. The relative rotation be-
magnetic propulsion system for specific application of active
tween magnet assembly and patient support permits adjustment
locomotion control of the endoscopic capsule in the GI tract.
of the positions of magnet, allowing magnet built-in capsule
Engineering design of the magnetic propelled endoscopic sys-
endoscope to rotate about its longitudinal axis. According to
tem is summarized in Fig. 2.
the position feedback information, the servo control unit gives
orders for driving the multiaxes of the magnetic propulsion sys-
A. Wireless Capsule Endoscope tem. Position control mode with pulse signal input is adopted
The developed endoscopic capsule is constructed by four and human–machine interface is developed for realizing semi-
main parts—illumination module, vision module, control- automatic control. During the operation procedure, input control
ling module, and wireless transmission module, as shown in parameters should be amended artificially for realizing the sta-
Fig. 2(a). The concept of the magnetic shell is proposed by ble movements of the capsule [see Fig. 2(b)].
Carpi et al. [24]–[26]. The microsystems integration of the cap- The proposed magnetic propulsion system has five coupling
sule endoscope is based on commercialized chips and devices axes that can realize capsule endoscope’s multidirection move-
available on the markets, thereby rendering the total costs of ment in the GI tract. Fig. 4 demonstrates the principle of capsule
2894 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 12, DECEMBER 2010

Fig. 5. (a) Geometry-optimized capsule. (b) Schematic of the external mag-

netic source.

of 8.82 g. The magnetic shell has a dimension of ø11 mm ×

ø6 mm × 10 mm, a weight of 6.95 g, and generate a maximum
magnetic field of 0.36 T at the external surface.

B. External Magnetic-Field Sources

Fig. 4. Locomotion principle of self-propelled endoscopic capsule. Electromagnets or permanent magnets can be mounted on
the magnet assembly of the proposed magnetic propulsion sys-
tem. In current approach, two groups of permanent Nd–Fe–B
endoscope’s locomotion mechanism. The different movements magnets are used to provide sufficient magnetic field for driving
of five axes are shown with numbers and corresponding arrow- endoscopic capsule [see Fig. 5(b)]. Each group includes a rect-
head in Fig. 4(a)–(c). Number 1 with red color indicates magnet angle Nd–Fe–B magnet and a cylindrical Nd–Fe–B magnet; the
assembly’s rotation about its longitudinal axis, number 2 with rectangle one has a length of 80 mm, a width of 60 mm, a height
green color shows two magnets’ linear motion in opposite direc- of 15 mm, and the cylindrical one has a diameter of 90 mm, a
tion in the coronal plane, number 3 with blue color represents height of 30 mm, and generated a field of 0.32 and 0.29 T at the
two magnets’ linear motion in opposite direction in the sagittal center of its surface, respectively.
plane, number 4 with purple color shows the patient support’s
linear movement in the lateral direction, and number 5 with or- C. Simulated Intestinal Tract
ange color indicates the patient support’s linear movement in the
longitudinal direction. By combining movements of five axes in Polyvinyl chloride (PVC) tube is used as the simulated in-
different permutation and sequence, we can navigate magnet testinal tract for experiments. The length and the diameter of
built-in capsule endoscope’s locomotion through the simulated the PVC tube is 1.6 m and 20 mm, respectively. The choice of
intestinal tract in multidirections. Four main movements of cap- the simulated intestinal tract is based on the geometry similar-
sule endoscope are shown in Fig. 4(d), the relative translation ity, and the dimension of the chosen PVC tube is approximate
between patient support and magnet assembly will steer cap- to the human’s small intestine, which has a typical diameter of
sule moving in the lateral or longitudinal direction in terms of 25–30 mm for adults [35].
patient support’s motion path; the opposite motion of magnets
in the coronal plane will drive capsule rotating about its verti- D. FE Analysis
cal axis, this motion changes capsule’s orientation, allowing to
realize capsule’s motion transition from the longitudinal path to FE methods are used to simulate magnetic-field distributions
the lateral path and vice versa. The capsule endoscope is also and mechanical properties of the proposed magnetic navigation
capable of rotating about its longitudinal axis by pivoting the system. Gauss meter (HT100G, HT Magnet Corporation) is used
magnet assembly in the magnet support, this movement is not to measure the actual values of generated magnetic field.
only useful for adjusting capsule’s orientation, but also neces-
sary for enlarging viewing angle of the built-in micro camera to E. Pressure Wave of Organ
improve diagnosis effect of GI endoscopy.
Nonlinear curve fitting is used for extracting pressure-wave
formula based on the in vivo measuring results in [36]. The
III. MATERIALS AND METHODS pressure wave turns out to be a Fourier series given by
A. Fabrication of the Magnet/PMMA Capsule Complex

N

Fig. 5(a) shows a geometry-optimized magnet/PMMA cap- a = A0 + [An sin nω(t − t0 ) + Bn cos nω(t − t0 )],
sule prototype based on orthogonal process design. The capsule n =1

has a diameter of 11 mm, a length of 26 mm, and a weight t ≥ t0 (1)

GAO et al.: DESIGN AND FABRICATION OF A MAGNETIC PROPULSION SYSTEM FOR SELF-PROPELLED CAPSULE ENDOSCOPE 2895

where a represents the pressure-wave formula (in kilopascals),

and t0 , N , ω, A0 , An , and Bn , n = 1, 2, . . . , N , are defined
constants.

IV. MODELING
A. Magnetic Propulsion Modeling
In a current free region [37], where
∇×H=0 (2)
it is possible to define the scalar magnetic potential Vm from the
relation
H = −∇Vm (3)
where H is the magnetic-field intensity.
This is analogous to the definition of the electric potential for
static electric fields. Using the constitutive relation between the
Fig. 6. Illustration of magnetic propulsion modeling.
magnetic-flux density B and magnetic field density H
B = μ0 (H + M) (4)
The equations for the balance of forces on the capsule now
where μ0 is the permeability of vacuum, and M is the magne- become
tization vector, together with the equation
∇ · T1 + fext = 0 (12)
∇·B=0 (5)
n1 (T2 − T1 ) + gext = 0. (13)
we can derive an equation for Vm
For computing the total force F on the capsule, these equa-
−∇ · (μ0 ∇Vm − μ0 M) = 0. (6) tions need to be integrated over the entire capsule and the cap-
sule/medium boundary
Along the boundaries far away from the magnet, the magnetic  
field should be tangential to the boundary as the flow lines should (∇ · T1 + fext )dV + (n1 (T2 − T1 ) + gext ) dS = 0.
form closed loops around the magnet. The natural boundary Ω1 ∂ Ω1
condition is (14)
Now, according to Gauss’ theorem
n · (μ0 ∇Vm − μ0 M) = n · B = 0 (7)  
where n represents the unit normal vector. Thus, the magnetic ∇ · T1 dV − n1 T1 dS = 0. (15)
Ω1 ∂ Ω1
field is made tangential to the boundary by a Neumann condition
on the potential. This means that the external force
 
Cauchy’s equation of continuum mechanics reads Fext = fext dV + gext dS (16)
2 Ω1 ∂ Ω1
d r
ρ = ∇ · T + fext (8) is needed to balance the term for the boundary integral of the
dt2
where ρ is the density, r denotes the coordinates of a material stress tensor in the surrounding medium

point, T is the stress tensor, and fext is an external volume
F= n1 T2 dS (17)
force. This is the equation solved in the structural mechanics ∂ Ω1
application modes for the special case of a linear elastic material.
In the stationary case (as shown in Fig. 6), there is no accel- to keep the capsule stationary, i.e.,
eration, and the equation representing the force balance is Fext + F = 0. (18)
0 = ∇ · T + fext . (9) To summarize, the total force F is computed as a boundary
On the boundary, the following equations apply: integral of the stress tensor in the medium on the outside of the
capsule. Note that to obtain this result, the contribution from the
n1 (T̄2 − T1 ) = 0 (10) medium pressure gradient has been neglected. This is equivalent
to assuming that ∇ · T2 = 0.
n1 T̄2 = n1 T2 + gext (11)
Due to limitations of carrying out in vivo experiments, sim-
where T1 and T2 represent the stress tensor in the capsule and ulated intestinal experiments were conducted in cut-open in-
the medium around the capsule in the GI tract. The external testine specimen instead of the living body. Therefore, fluidic
boundary force gext can represent the reaction force from the medium of the intestine is approximated by air. However, in
wall of the intestine that the capsule is attached to. order to estimate the actual stress tensor of the fluidic medium
2896 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 12, DECEMBER 2010

on the outside of the capsule endoscope, the effect due to the

fluidic medium of the intestine should be considered.
For air, the stress tensor [38], [39]
 
ε0 1 1
T2 = −pI − E·E+ B · B I + ε0 EET + BBT
2 2μ0 μ0
(19)
where p is the air pressure, ε0 is the permittivity of vacuum, I is
the identity 3×3 tensor (or matrix), and electric field intensity E
and magnetic-flux density B are 3×1 vectors. In this expression
of the stress tensor, air is considered to be nonpolarizable and
nonmagnetizable. When air is approximated by vacuum, p = 0.
This expression of the stress tensor is also known as the Maxwell
stress tensor.
Using the fact that, for air, electric flux density D = ε0 E and Fig. 7. Frictional resistance modeling of capsule (redrawn based on [40]).
B = μ0 H, the expression for the stress tensor can be written as
 
1 1
T2 = −pI − E · D + H · B I + EDT + HBT . (20)
2 2
The equation for the balance of forces becomes
  
1 1
0 =∇ · −pI − E·D+ H·B I
2 2

+ EDT + HBT + fext . (21)
Consider again the case of a capsule endoscope surrounded
by air. To compute the total force, the projection of the stress
tensor on the outside of the capsule surface is needed,
 
1 1 Fig. 8. Magnetic-flux density as a function of distance.
n1 T2 = − pn1 − E · D + H · B n1
2 2
+ (n1 · E)DT + (n1 · H)BT (22) 3) The deformed shape of the GI tract is equal to that of the
capsule’s contact region.
where n1 is the surface normal, a 1×3 vector, pointing out from 4) There is no inclination of the capsule relative to the axis
the capsule. This expression can be used directly in the boundary of the small intestine.
integral of the stress tensor for computing the total force F on The total frictional force can be expressed as follows [40]:
the endoscopic capsule.
F = FW + F V + F P (23)
B. Frictional Resistance Modeling where FW is the frictional force related to the capsule weight,
FV is the frictional force induced by the stress due to the vis-
Given the facts that the power consumption and position con-
coelastic deformation of the intestine tract, and FP is the fric-
trol of the endoscopic capsule are largely affected by the fric-
tional force induced by the stress due to the pressure of the
tional resistance of the endoscope inside the GI tract and the
external organs. The friction component FP due to the pressure
viscoelastic characteristics of the internal organs, fundamental
wave is not considered in [40]; however, in case of pressure of
understanding of these characteristics is of vital importance;
external organs, the friction coefficient may be dramatically in-
and thus, modeling and simulation of capsule’s mechanical lo-
creased. In the present paper, three major components (FW , FV ,
comotion in GI tract are definitely necessary for prediction of
and FP ) are considered in the FE simulations. FE analysis was
frictional forces of the endoscopic capsule moving inside the
performed by using ABAQUS and the results will be discussed
body [40].
in the following.
Kim et al. established contact mechanics model for analyzing
the frictional resistance of a capsule endoscope moving inside
the body, as shown in Fig. 7 [40]. For the model development, V. RESULTS AND DISCUSSION
the following assumptions are made. Fig. 8 shows the relation between the magnetic-flux density
1) The GI tract consists of an isotropic material and the and the distance. The black line plots the simulation results,
small intestine deforms symmetrically toward its radial whereas the circle plots the experiment measurement values.
direction. The experimental data match well with the FE analysis, demon-
2) The volume and the wall thickness of the GI tract are strating that our electromagnetic model is appropriate for an-
constant. alyzing the mechanical properties of the proposed propulsion
GAO et al.: DESIGN AND FABRICATION OF A MAGNETIC PROPULSION SYSTEM FOR SELF-PROPELLED CAPSULE ENDOSCOPE 2897

methacrylate (PMMA) capsule. The triangle symbol shows the

relationships of the capsule’s locomotion speed versus the fric-
tional resistance force of the porcine intestine specimen. The
frictional resistance force was measured by Wang and Meng
through a comprehensive experimental study [41]. There are 15
capsule prototypes, and both closed and open intestine experi-
ments were conducted in Wang’s research. We use data of no.
4 in close porcine intestine in terms of three considerations:
the dimension of no. 4 capsule prototypes in [41] experiments
is same to our capsule (ø11 mm × 26 mm); the locomotion
speed range (0.5–8 mm/s) falls within the control capability of
the developed magnetic propulsion system; the status of close
intestine is reflected by the FE model and approximates to the
practical situation. In order to overcome the frictional resistance
Fig. 9. Visualization of the solved magnetic field. of the intestine, the magnitude of distance between external and
internal magnets must be adjusted to an appropriate value for
generating enough driving force, which can be obtained in terms
of square plot line. As we can see, permanent-magnet sources
with a surface magnetic-flux density of 0.34–0.5 T is appropriate
for driving the magnet/PMMA capsule complex, and the critical
distance between the capsule and the external magnetic source
is about 12–15 cm. Therefore, operation distance between two
groups of magnetic sources can be approximately 24–30 cm. By
further optimization of the proposed propulsion system, an op-
eration area with dimensions of 40 cm × 40 cm can be achieved,
which is appropriate for serving the patients with various body
conformations (see Table I).
Although the magnetic analysis indicates the feasibilities of
the proposed magnetic propulsion system for driving the endo-
scopic capsule moving inside the intestine, the actual intestinal
environment is much more complicated due to the viscoelas-
Fig. 10. Comparison between the magnetic force and the friction force (ex- tic feature, pressure wave of the external organs, and sophisti-
periment results are from [41]). cated surface topography of the intestinal tract. We performed
FE simulation by using ABAQUS 6.8 and HYPERMESH 8.0.
system. As we can see, with the increase of the distance be- Viscoelastic characteristics of the intestine, pressure wave, and
tween magnetic sources and built-in magnets, the strength of contact effect between the capsule and the small bowel are
the magnetic-flux density drops rapidly. Thus, tradeoffs between considered. In addition, an orthogonal process design with five
scope of operation area and strength of magnetic field should factors and two levels is carried out for design optimization and
be considered for providing optimal control parameters of the analyzing main effects and interaction for frictional force of the
magnetic propulsion system. capsule’s locomotion inside the intestinal tract (parameters and
Fig. 9 is a visualization of the solved magnetic field. The values are listed in Table II). Fig. 11 shows the geometry model
streamlines show the lines of magnetic force, and the back- of endoscopic capsule’s locomotion in the intestinal tract. Five
ground color shows the distribution of the magnetic field. The factors—radius of the capsule, radius of curvature, length of
magnetic field is stronger when the magnetic force lines be- the capsule, velocity of the capsule, and frictional coefficient of
come denser. A maximum intensity of 0.34 T at the surface is the intestinal tract are considered for performing the orthogonal
obtained. The streamline in the magnetic field maps indicates process design.
that the relative orientation between the inner and outer magnets A two-dimension axisymmetric FE model is established for
is critical for an accurate and controlled navigation. The appear- conducting FE analysis. CAX4 H element (four-node bilinear,
ance of capsule could affect the magnetic circuit. A maximum hybrid with constant pressure) is chosen because it can avoid
control force can be obtained when the magnetization plane of shear locking and can accurately reflect large deformations due
the capsule is orthogonal to the surface of the external magnetic to the contact effect. A contact pair is defined that embodies
sources. surface elements of the capsule and the small bowel. The small
Fig. 10 illustrates the comparison between the magnetic driv- intestine is defined as the slave surface and the capsule as the
ing force and the frictional resistance force. A square sym- master surface in terms of the physical feature of corresponding
bol plots the relationships of the distance between external and materials. Stress relaxation material data are based on experi-
internal magnets versus the value of magnetic force exerted mental measurements from [42]. Fig. 12 shows the distributions
on the built-in Nd–Fe–B magnets of the magnet/polymethyl of von Mises stress that are applied along the capsule of sample
2898 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 12, DECEMBER 2010

TABLE II
ORTHOGONAL ARRAY OF AVERAGE FORCE UNDER FIVE FACTORS

Fig. 13. Fitting curve of the pressure (absolute pressure) of the external organs.

TABLE III
Fig. 11. Geometry model of capsule’s locomotion in the intestinal tract. SUMMARY OF COEFFICIENTS

Fig. 12. FE simulations of von Mises stress distributions allied along the
capsule moving inside the intestinal tract (in megapascals).

8. Contact effects, pressure wave as well as viscoelastic char-

acteristics can be found. The fitting curve of pressure wave is
shown in Fig. 13. Table III summarizes the coefficients in (2).
It should be noted that A0 , An , and Bn , n = 1, 2, . . . , 8, have a
unit of kilopascal. The pressure formula is embodied in the FE
simulations.
FE simulations of the frictional force variation of eight sam-
ples are shown in Fig. 14. FE results are similar to the experi-
mental measurements from [43], indicating that we can use the
FE simulations to obtain important control parameters for the
proposed magnetic propulsion system. Scenarios with (samples Fig. 14. Frictional force variation of eight samples with different geometric
and physical parameters (pressure wave is exerted on the intestinal wall with
1, 3, 5, and 7) and without (samples 2, 4, 6, and 8) pressure-wave position of 30–40 mm).
action are performed for the purpose of visualizing the effect
of the organ pressure wave. The result shows that the resistant
force is significantly affected by the external pressure wave; in the intestine. It can be found that the frictional coefficient,
it indicates that design optimization is necessary for a robust the radius of the capsule, and the length of the capsule con-
control and appropriate propulsion. Based on the FE analysis, tribute greatly to the magnitude of the frictional force. With the
an orthogonal process design with five factors and two levels is increase of the frictional coefficient, and the radius and length
conducted by using MINITAB 15. Fig. 15 shows the main ef- of the capsule, the frictional resistance force increases signif-
fects plot for average frictional force of the capsule’s locomotion icantly. The radius of curvature and the capsule’s velocity are
GAO et al.: DESIGN AND FABRICATION OF A MAGNETIC PROPULSION SYSTEM FOR SELF-PROPELLED CAPSULE ENDOSCOPE 2899

Fig. 15. Main effects analysis of frictional force with five factors.

Fig. 16. Interaction effects analysis of frictional force with five factors.

Fig. 18. Movement of the capsule on a simulated intestinal tract.

frictional coefficient is favorable for reducing the frictional re-

Fig. 17. Magnetic propulsion system for capsule endoscopy.
sistance, and thus, facilitating the movement of the endoscopic
capsule inside the intestinal tract.
secondary causes in this case. Fig. 16 indicates the interaction According to the results of the orthogonal process design, a
effects of the frictional force of capsule moving inside the small geometry-optimized capsule is fabricated, as shown in Fig. 5(a).
bowel. There are no or little interactions between R and ρ, R and A tube-shaped Nd–Fe–B magnet (11 mm ø (6 mm ø) × 10 mm),
V, ρ and L, ρ and f, L and V, and V and F. However, interactions which is magnetized in the direction of its diameter, is con-
exist between R and L, R and f, ρ and V, and L and f. We can see nected with two hemispheroids made of PMMA. The front
that combinations of high frictional coefficient and high radius, hemispheroid and the rear one have a radius of 5.5 mm and
or high length result in a great increase of the frictional force. a radius of curvature of 2 and 1.5 mm, respectively.
Therefore, design of endoscopic capsule with smaller radius and Proposed magnetic propulsion system for intestinal en-
length as well as suitable surface material that can reduce the doscopy is developed, as shown in Fig. 17. Experiments of
2900 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 12, DECEMBER 2010

TABLE IV
SUMMARY OF In Vitro TESTS

The proposed approach suggests the feasibilities of steering

the endoscopic capsule’s location and orientation by the devel-
oped magnetic propulsion system. However, further investiga-
tions should be carried out for improving the current system in
two aspects: 1) external image visualization unit is necessary
as the supplementary component of the magnetic propulsion
system for providing position information of the capsule. In the
current system, eyes serve as the image visualization unit for the
servo control unit to perform external manipulation. 2) Fluidic
medium of the GI tract and the closed forms of the human body
should be taken into consideration for simulating the actual en-
vironment of the capsule inside the body cavity under external
magnetic equipment.
Fig. 19. Simulated intestinal tract (a) PVC tubes. (b) PVC tubes with intestinal
layers. VI. CONCLUSION
A novel magnetic propulsion system was proposed and manu-
factured for controlling endoscopic capsule’s movements in the
motion control are conducted in a transparent PVC tubes and intestinal tract. The development of the proposed system was
are presented by the video frames reported in Fig. 18. It shows pursued through a combined program of analysis, numerical
a combined movement of translation and rotation of the capsule simulations, and experiments. Locomotion analysis and sys-
(marked by red color). Due to the visible external manipula- tem design were performed and magnetic-analysis model was
tion of magnetic navigation system, the capsule can move lin- established for calculating the magnetic-field distributions and
early in both longitudinal and lateral directions, rotate about its analyzing the mechanical properties of the proposed propulsion
longitudinal and lateral axes, and pivot about its vertical axis. system. Based on an orthogonal process design, FE simulations
Fig. 18(a1)–(f1) shows the location and the moving direction of were carried out and a geometry-optimized capsule was fab-
the capsule at different time in the simulated intestinal tract. The ricated. Simulated intestinal tract experiments were conducted,
position and orientation of the magnet assembly and the patient indicating that the proposed technique is suitable for steering the
support inducing these movements are reported in the related endoscopic capsule for performing noninvasive GI endoscopy.
frames Fig. 18(a2)–(f2). Further research is currently in progress for visualizing the lo-
The artificial path was patterned in terms of the tortuous comotion control by ways of ultrasonic image.
feature of the intestine. In vitro experiments are carried out in
the transparent PVC tube with and without intestinal layer. The ACKNOWLEDGMENT
dimension of the simulated intestine is ø20 cm × 160 cm and it The authors greatly appreciate the help rendered by Dr. D. Xie
includes six regions, which represent different combination of and Engineer Y. Xu of the School of Mechanical Engineering,
straight and curved paths (see Fig. 19). Three strictures are set up Huazhong University of Science and Technology for their help
to simulate the physiological characteristics of the small bowel. in manufacturing this magnetic propulsion system. The authors
Physiological saline is sprayed on the porcine small intestine also would like to thank Prof. D. Zhang of the California State
to avoid drying of the tissue. The locomotion performance of University for his help in finite-element modeling.
the magnetic propulsion system is recorded for each region of
the simulated tract. Table IV summarizes the position and the
REFERENCES
velocity information of the capsule prototype. The total time of
the capsule for going through the 1.6-m-long tract is about 53.09 [1] G. Iddan, G. Meron, A. Glukhovsky, and P. Swain, “Wireless capsule
endoscopy,” Nature, vol. 405, pp. 417–417, May 2000.
s for PVC tubes [see Fig. 19(a)] and 148.78 s for intestine [see [2] M. Hashizume and K. Tsugawa, “Robotic surgery and cancer: the present
Fig. 19(b)], respectively. The average velocity is about 30.14 state, problems and future vision,” Jpn. J. Clin. Oncol., vol. 34, pp. 227–
and 10.75 mm/s, respectively. 237, May 2004.
GAO et al.: DESIGN AND FABRICATION OF A MAGNETIC PROPULSION SYSTEM FOR SELF-PROPELLED CAPSULE ENDOSCOPE 2901

[3] C. Mc Caffrey, O. Chevalerias, C. O’Mathuna, and K. Twomey, [25] F. Carpi, S. Galbiati, and A. Carpi, “Controlled navigation of endoscopic
“Swallowable-capsule technology,” IEEE Pervasive Comput., vol. 7, capsules: Concept and preliminary experimental investigations,” IEEE
no. 1, pp. 23–29, Jan./Mar. 2008. Trans. Biomed. Eng., vol. 54, no. 11, pp. 2028–2036, Nov. 2007.
[4] I. Kassim, W. S. Ng, G. Feng, S. J. Phee, P. Dario, and C. A. Mosse, [26] F. Carpi and C. Pappone, “Magnetic maneuvering of endoscopic capsules
“Review of locomotion techniques for robotic colonoscopy,” in Proc. 20th by means of a robotic navigation system,” IEEE Trans. Biomed. Eng.,
IEEE Int. Conf. Robot. Autom. (ICRA), Taipei, Taiwan, 2003, pp. 1086– vol. 56, no. 5, pp. 1482–1490, May 2009.
1091. [27] J. Peirs, D. Reynaerts, and H. Van Brussel, “Design of miniature parallel
[5] B. S. Lewis and P. Swain, “Capsule endoscopy in the evaluation of pa- manipulators for integration in a self-propelling endoscope,” in Proc.
tients with suspected small intestinal bleeding: Results of a pilot study,” Eurosensors XIII Meeting, The Hague, Netherlands, 1999, pp. 409–417.
Gastrointest. Endosc., vol. 56, pp. 349–353, Sep. 2002. [28] V. R. C. Kode and M. C. Cavusoglu, “Design and characterization of
[6] M. Mylonaki, A. Fritscher-Ravens, and P. Swain, “Wireless capsule en- a novel hybrid actuator using shape memory alloy and DC micromotor
doscopy: A comparison with push enteroscopy in patients with gas- for minimally invasive surgery applications,” in Proc. 1st IEEE RAS-
troscopy and colonoscopy negative gastrointestinal bleeding,” Gut, EMBS Int. Conf. Biomed. Robot. Biomechatron. (BioRob), Pisa, Italy,
vol. 52, pp. 1122–1126, Aug. 2003. 2006, pp. 455–464.
[7] M. Y. Gao, C. Z. Hu, Z. Z. Chen, S. Liu, and H. H. Zhang, “Finite- [29] H. G. Sim, S. K. H. Yip, and C. W. S. Cheng, “Equipment and technology
difference modeling of micromachine for use in gastrointestinal en- in surgical robotics,” World J. Urol., vol. 24, pp. 128–135, Jun. 2006.
doscopy,” IEEE Trans. Biomed. Eng., vol. 56, no. 10, pp. 2413–2419, [30] D. E. Koditschek, R. J. Full, and M. Buehler, “Mechanical aspects of
Oct. 2009. legged locomotion control,” Arthropod Struct. Dev., vol. 33, pp. 251–
[8] A. Menciassi, M. Quirini, and P. Dario, “Microrobotics for future gas- 272, Jul. 2004.
trointestinal endoscopy,” Minim. Invasive Ther. Allied Technol., vol. 16, [31] C. Pawashe, S. Floyd, and M. Sitti, “Modeling and experimental charac-
pp. 91–100, 2007. terization of an untethered magnetic micro-robot,” in Proc. 4th Robot. Sci.
[9] A. Moglia, A. Menciassi, M. O. Schurr, and P. Dario, “Wireless capsule Syst. Conf., Zurich, Switzerland, 2008, pp. 1077–1094.
endoscopy: From diagnostic devices to multipurpose robotic systems,” [32] S. Kumar, I. M. Kassim, and V. K. Asari, “Design of a vision-guided
Biomed. Microdevices, vol. 9, pp. 235–243, Apr. 2007. microrobotic colonoscopy system,” Adv. Robot., vol. 14, pp. 87–104,
[10] M. Delvaux and G. Gay, “Capsule endoscopy: Technique and indications,” 2000.
Best Pract. Res. Clin. Gastroenterol., vol. 22, pp. 813–837, 2008. [33] K. B. Yesin, K. Vollmers, and B. J. Nelson, “Modeling and control of
[11] T. Nakamura and A. Teran, “Capsule endoscopy: Past, present, and future,” untethered biomicrorobots in a fluidic environment using electromagnetic
J. Gastroenterol., vol. 43, pp. 93–99, 2008. fields,” in Proc. 9th Int. Symp. Exp. Robot. (ISER), Singapore, 2004,
[12] K. Ishiyama, K. I. Arai, M. Sendoh, and A. Yamazaki, “Spiral-type micro- pp. 527–536.
machine for medical applications,” in Proc. Int. Symp. Micromechatron. [34] New world encyclopedia. (Nov. 10, 2007). [Online]. Available:
Human Sci., Nagoya, Japan, 2000, pp. 65–69. http://www.newworldencyclopedia.org/entry/Gastrointestinal_tract
[13] M. Sendoh, K. Ishiyama, and K. I. Arai, “Fabrication of magnetic actuator [35] Wikipedia. (2006). [Online]. Available: http://en.wikipedia.org/wiki/
for use in a capsule endoscope,” in Proc. Int. Magn. Conf., Boston, MA, Small_intestine
2003, pp. 3232–3234. [36] S. Park, J. Park, H. Park, S. Park, C. Jee, S. Park, and B. Kim, “Multi-
[14] A. Yamazaki, M. Sendoh, K. Ishiyama, T. Hayase, and K. I. Arai, “Three- functional capsule endoscope for gastro-intestinal tract,” presented at the
dimensional analysis of swimming properties of a spiral-type magnetic SICE-ICASE Int. Joint Conf., Busan, Korea, Oct. 2006.
micro-machine,” Sens. Actuators A—Phys., vol. 105, pp. 103–108, Jun. [37] R. P. Feynman, R. B. Leighton, and M. Sands, The Feynman Lectures on
2003. Physics,. vol. 2, Reading, MA: Addison-Wesley, 1964, ch. 14.
[15] A. Chiba, M. Sendoh, K. Ishiyama, and K. I. Arai, “Moving of a magnetic [38] A. Kovetz, The Principles of Electromagnetic Theory. Cambridge, U.K.:
actuator for a capsule endoscope in the intestine of a pig,” J. Magn. Soc. Cambridge Univ. Press, 1990.
Jpn., vol. 29, pp. 343–346, 2005. [39] O. Wilson, Introduction to Theory and Design of Sonar Transducers.
[16] A. Chiba, M. Sendoh, K. Ishiyama, Y. Suda, K. I. Arai, T. Komaru, Los Altos, CA: Peninsula, 1988.
and K. Shirato, “Colon endoscope navigation by magnetic actuator and [40] J. S. Kim, I. H. Sung, Y. T. Kim, D. E. Kim, and Y. H. Jang, “Analytical
intestine observations,” J. Magn. Soc. Jpn., vol. 28, pp. 433–436, model development for the prediction of the frictional resistance of a
2004. capsule endoscope inside an intestine,” in Proc. Inst. Mech. Eng. Part
[17] B. Kim, S. Lee, J. H. Park, and J. O. Park, “Design and fabrication of a H—J. Eng. Med., Nov. 2007, vol. 221, pp. 837–845.
locomotive mechanism for capsule-type endoscopes using shape memory [41] X. Wang and M. Q. H. Meng, “An experimental study of resistant proper-
alloys (SMAs),” IEEE-ASME Trans. Mechatron., vol. 10, no. 1, pp. 77–86, ties of the small intestine for an active capsule endoscope,” in Proc. Inst.
Feb. 2005. Mech. Eng. Part H—J. Eng. Med., 2010, vol. 224, pp. 107–118.
[18] C. A. Mosse, T. N. Mills, M. N. Appleyard, S. S. Kadirkamanathan, and [42] J. S. Kim, I. H. Sung, Y. T. Kim, E. Y. Kwon, D. E. Kim, and Y. H. Jang,
C. P. Swain, “Electrical stimulation for propelling endoscopes,” in Proc. “Experimental investigation of frictional and viscoelastic properties of
Digestive Disease Week/101st Annu. Meeting Amer. Gastroenterol. Assoc., intestine for microendoscope application,” Tribol. Lett., vol. 22, pp. 143–
San Diego, CA, 2000, pp. 79–83. 149, May 2006.
[19] L. Phee, D. Accoto, A. Menciassi, C. Stefanini, M. C. Carrozza, and [43] N. K. Baek, I. H. Sung, and D. E. Kim, “Frictional resistance characteris-
P. Dario, “Analysis and development of locomotion devices for the gas- tics of a capsule inside the intestine for microendoscope design,” in Proc.
trointestinal tract,” IEEE Trans. Biomed. Eng., vol. 49, no. 6, pp. 613–616, Inst. Mech. Eng. Part H—J. Eng. Med., May 2004, vol. 218, pp. 193–201.
Jun. 2002.
[20] P. Dario, P. Ciarletta, A. Menciassi, and B. Kim, “Modelling and exper-
imental validation of the locomotion of endoscopic robots in the colon,”
in Proc. 8th Int. Symp. Exp. Robot. (ISER 2002), Sant Angelo, Italy,
pp. 445–453.
[21] A. Menciassi, A. Moglia, S. Gorini, G. Pernorio, C. Stefanini, and P. Dario,
“Shape memory alloy clamping devices of a capsule for monitoring tasks
in the gastrointestinal tract,” J. Micromech. Microeng., vol. 15, pp. 2045–
2055, Nov. 2005.
[22] M. Quirini, A. Menciassi, S. Scapellato, C. Stefanini, and P. Dario, “Design Mingyuan Gao (S’10) was born in Yinchuan, China,
and fabrication of a motor legged capsule for the active exploration of the in 1985. He received the M.S. degree in mechanical
gastrointestinal tract,” IEEE ASME Trans. Mechatron., vol. 13, no. 2, manufacturing and automation from Huazhong Uni-
pp. 169–179, Apr. 2008. versity of Science and Technology, Wuhan, China, in
[23] K. D. Wang, G. Z. Yan, G. Y. Ma, and D. D. Ye, “An earthworm-like robotic 2010.
endoscope system for human intestine: design, analysis, and experiment,” His current research interests include biomedical
Ann. Biomed. Eng., vol. 37, pp. 210–221, Jan. 2009. equipment, biomechanical modeling, and microsys-
[24] F. Carpi, S. Galbiati, and A. Carpi, “Magnetic shells for gastrointesti- tems packaging.
nal endoscopic capsules as a means to control their motion,” Biomed. Mr. Gao received the 2009 GE Foundation TECH
Pharmacother., vol. 60, pp. 370–374, Sep. 2006. Award—first place award.
2902 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 12, DECEMBER 2010

Chengzhi Hu received the B.S. degree in measur- Honghai Zhang received the Ph.D. degree in me-
ing and control technology and instrumentation and chanical manufacturing engineering from Huazhong
the M.S. degree in mechanical manufacturing and au- University of Science and Technology, Wuhan,
tomation from Huazhong University of Science and China, in 1988.
Technology, Wuhan, China, in 2008 and 2010, re- He is currently a Professor at the National Lab-
spectively. He is currently working toward the Ph.D. oratory for Optoelectronics, Huazhong University of
degree at Nagoya University, Nagoya, Japan. Science and Technology. His research interests in-
His research interests include simulations and clude micro/nanomanufacturing technology. He has
analysis of electromagnetics, automation and micro- developed a number of micromanufacturing equip-
robotics, microsystems packaging technology, and ments, such as multifunctionally metrological scan-
active control of capsule endoscopy. ning probe microscopy, nanoimprinting equipment,
microjetting platform, and equipment for measurement and control of surface
microcharacteristics in precision process.

Sheng Liu (M’93–SM’09) received the B.S. and

M.S. degrees from Nanjing University of Aeronau-
tics and Astronautics, Nanjing, China, in 1983 and
Zhenzhi Chen received the B.S. degree in mechan-
1986, respectively, and the Ph.D. degree from Stan-
ical science and engineering from the Harbin Insti-
tute of Technology, Harbin, China, in 2008. Since ford University, Stanford, CA, in 1992.
He is currently with the School of Mechanical
August 2008, he has been working toward the Post-
Science and Engineering, and the National Labora-
graduate degree at the Institute of Microsystems,
tory for Optoelectronics, Wuhan, China. His research
Huazhong University of Science and Technology,
interests include microsystems (MS)/nano-electro-
Wuhan, China.
mechanical systems (NEMS), LED design and manu-
His research interests include design and model-
facturing, system packaging and integration, reliabil-
ing for self-propelled robotic endoscopy system.
ity, smart materials and composites, and mechanics of materials and structures.
Dr. Liu is a Fellow of the American Society of Mechanical Engineers.