Locomotion And Steering Aspects In Automation Of colonoscopy ...

2 downloads 0 Views 2MB Size Report
Automation of colonoscopy can be broken down into two aspects: locomo- tion and steerable distal end. The prob- lems of locomotion are concerned with.
locomotion and Steering Aspects in Automation of Colonoscopy Purt One: A literature Review iseases of the colon, rectum, and anus D are common in all countries. Poor eating habits, irregular toilet practices, and a harmful lifestyle combine with genetic factors to affect the large bowel, resulting in irritable bowel syndrome, diverticulosis, colorectal cancer, rectal prolapse, hemorrhoids, and inflammatory bowel disease. Almost all large bowel symptoms demand a full evaluation of the colon and rectum. Historically, many instruments have been devised to help inspect the rectum and colon. But it was not until 1963 that the first flexible colonoscope and sigmoidoscope were invented. These instruments allowed the endoscopist to inspect the entire colon and to treat colonic diseases without the need for open surgery. Medical technology is changingrapidly with devices such as stereo-endoscopes, virtual reality, telepresence surgery, and robotics for use in colonoscopy. Colonoscopic simulators, threedimensionalcolonoscopicimagers, and ultrasonic scanners are already developed, facilitated by more powerful, desktop computers. Within the next decade, it is probable that colonoscopy will utilize a miniatureendoscopic“robot” that walks or glides up the colon, blasting polyps with a laser cannon while supplying spectroscopic tissue informationfor instant analysis, and marking the site of lesions on a holographic image of the patient’s colon. Automation of colonoscopy can be broken down into two aspects: locomotion and steerable distal end. The problems of locomotion are concerned with activation systems and control sequencing of actuator elements within an environment that is slippery and has a highly varied anatomy. Steerable tips have difficulties associated with the need for remote activation and the need to negotiate tight radii without penetrating the colon wall. This literature survey gives a critical review of research in these areas. November/Detember 1997

S.J. Phee’, W.S. Ng’, I.M. Chen’, F. Seow-Choen2, B.L. Davies3 ’Nanyan Technological University, Singapore ‘Singapore General Hospital 31mperial College, UK

The Conventional Colonoscope and Its Drawbacks Light transmission by flexible fiberoptic bundles was pioneered in the 1950s, and was almost immediately applied to the flexible gastroscopy. The flexible sigmoidoscope was first used clinically in 1963, and the first series of patients was

1. (a) The conventional colonoscope; (b) control unit (close up). (Adapted from [2].) IEEE ENGINEERING IN MEDICINE AND BIOLOGY

0739.51 75/97/$10.0001997

85

Suction

-,

r Air

Port

Standard Video Scope

-

,

Water

Umbilical Cord

Suction

Torqueing of the colonoscope is also required in such a situation. A detailed examination of the mucosa is performed both as the colonoscope is introduced and when it is slowly removed from the caecum. If the colonoscope is kept free of loops, the tip responds well and the exarmnation is facilitated. This is especially true if a therapeutic procedure (such as polypectomy) is to be undertaken, because large, redundant loops of the colonoscope can make control of the tip very difficult. Drawbacks of Conventional Colonoscopy The invention of the colonoscope was a quantum leap toward noninvasive surgery. It has benefited humans in many

2. Basic colonoscope design. (Adapted from [2].)

presented to the American Society for Gastrointestinal Endoscopy (ASGE) in May 1967 [2].Colonoscopes are theresult of a combination of flexible fiber-optic technology with the ability to make an instrument that can be advanced along the entire colon. In 1969, the Olympus Corp. built the first commercial colonoscope [2].Since then, colonoscopy has become a routine procedure in many hospitals all over the world.

end-on view of the tip shows a cross-section of the insertion tube contents: biopsylsuction channel, aidwater channel, light guides, and objective lens (Fig. 3)

The Basic Design The conventional colonoscope consists of an umbilical cord that attaches the scope to the light source, a controlhand grip unit, and an insertion tube of variable length (Fig. 1). Running through the umbilical cord, control unit, and insertion tube are a suction channel, an aidwater channel, and fiber-optic bundles for light transmission. The suction and aidwater channels are controlled by valves set into the control unit (Fig. 2). In the insertion tube, a biopsy channel and control wires from the angulation knobs on the control unit to the in-

The Colonoscopy Examination Colonoscopy is one of the most technically demanding endoscopic examinations and is very unpopular with patients. It is an art to coax an almost 2 m long flexible tube around a tortuous colon while causing minimal discomfort and yet performing a thorough examination. Most experienced colonoscopists use similar endoscopic techniques. Air is pumped into the colon to distend it and a d insertion. However, excessive air and overdistention should be avoided. The insertion pressure on the device must be gentle to avoid stretching the colonic wall or mesentery, which can cause pain or perforation. The colonoscope is advanced, by the pushing action of the endoscopist’s hand, to the caecum under direct vision. The lumen should be kept always in view SO that little or none of the operation is per-

strument tip are added The biopsy port,

formed blind A variety of “in-and-out’’

3. End-on view of colonoscope tip. (Adapted from [2].)

usually sited high on the insertion tube, the colonoscope, keeping the

on the manufactur plastic sheath surround distal end is designed for deflection under control of the guidewires, and the tip is contained by a screw-on cap. Due to its rigidity, the distal end is usually slightly thicker than the rest of the insertion tube. An 86

minimize looping and discomfort. In a difficult colon, special maneuvers (such as reducing the “alpha loop” in the sigmoid colon) are used to pass the sharply angu- 4. Loops in the insertion tube. (Adapted lated sigmoiddescending colon junction. from [ Z ] . ) IEEE ENGINEERING IN MEDICINE AND BIOLOGY

November/Detember 1997

ways not thought possible just a few decades ago. However, there is still room for further improvement. The basic act of maneuvering the colonoscopic tip around the many bends of the colon requires many years of practice and training. During the operation, the lumen may disappear from the surgeon’s sight, leading to a “red-out” when the tip is against the colonic wall, or worse; a “white-out,” when the tip stretches the colonic wall. When this happens, an inexperienced endoscopist may be disorientated and have difficulty looking for the lumen. Colonic perforation may occur. Furthermore, abrupt movements of the scope may result in tearing of the inner wall of the colon, which may in turn lead to excessive bleeding. The present colonoscope also requires the endoscopist to hold the control device with one hand leaving only one hand to push or pull the insertion tube. Too much torqueing of the insertion tube may result in loops, which may complicate matters further (Fig. 4). Besides being cumbersome, holding up the control device for prolonged periods of time is tiring for the endoscopist. The present colonoscopy procedure depends very much on the skills of the surgeon. A more experienced endoscopist will perform a more thorough, less painful operation in a shorter time than an inexperienced endoscopist. A skilled endoscopist will normally have little problems traversing the colonoscope right up to the caecum of a “normal” colon. However, there will be difficulties traversing the colonoscope through some “difficult” colons. This happens when encountering very acute or fixed bends. Further pushing of the colonoscope at this point will only distend the walls of the distal colon. Distortion of the colonic shape and profile due to previous surgery may add to this problem. To overcome these technical difficulties, a colonoscope with a traversing mechanism at its distal end could solve many locomotion problems, unlike the pushing technique used with conventional colonoscopes. Polyp removal from the colon walls can also cause difficulties. If there are a few polyps present, the surgeon will have to remove them one at a time. If the polyps are large, the colonoscope may have to be reinserted to look for the next polyp. Small polyps may often be retrieved with November/Detember 1997

5. Experimental model of an active endoscope. (A.daptedfrom [3].)

Intermediate Flanges

SHA Coil SpringsSpinal coil Spring

[x-arrayl

Cooling Water Tubes

6. Inner structure of an active endoscope. (Adapted from [3].)

a polyp trap. A biopsy net may be used to collect polyps and reduce this problem. However, one cannot then distinguish which location in the colon a particular polyp comes from. It is important for the endoscopist to know which part of the colon a particular polyp is removed from if subsequent therapy becomes necessary after histological examination. Apart from technical difficulties experienced by the doctor, the trauma and pain experienced by the patient may be considerable. The need for insufflating air into the colon in order to open its normally collapsed lumen causes discomfort for the patient. Furthermore, an inexperienced endoscopist may cause additional pain by using the wrong technique or too much unnecessary force. Even experienced endoscopists may cause pain if the patient is anxious, suffering from irritable bowel syndrome, or if the colon is fixed by adhesion or disease. IEEE ENGINEERING IN MEDICINE AND BIOLOGY

\Nhy Automate Colonostopy? The problems mentioned above may be solved by automating colonoscopy. The following are some advantagesof this approach: Automation removes the need for experience and skills of the operator. This means that a patient will have the same treatment,in terms of time taken and comfort, regardless of the endoscopist performing the examination. Training of an endoscopist may be reduced to learning treatments for abnormalities found on the intestinal walls, without the need to perfect the manual skills required to use a conventional colonoscope. With an automated procedure, more operations can be done by one surgeon, since only diagnosis will be required, thus reducing costs. Reduced trauma and discomfort for patients. Reduction of postoperative complications and hospitalization.

. .

.

. .

87

By automating the colonoscopy procedure, the skills of the surgeon will no longer be the dependent factor. Instead, any movements of the colonoscopewill be controlled by computers, which are faster and more precise and consistent than the human being. The endoscopist, however, must be present at all times to guide the machine to do its job. It is still the person who will decide every move the machine makes and who will take over when there are uncertainties or in case of an emergency.

Automation of Colonos~o~y f automating colonoscopy can be broken down into two distinct parts; locomotion and manipulation of the steerable distal end. Locomotion includes de-

-

signing a robot that can propel itself from the anus right up to the caecum, without damaging the delicate colon walls. In addition, the robot must have the ability to carry along optical fibers, surgical tools, and other instruments required in a colonoscopic procedure. Traversing along the colon is not enough. A steerable distal end must be designed to point the tip of the robot in a desired direction. This will allow the endoscopist to perform a more thorough examination by “looking around,” and also guide and position a surgical tool if surgery is required.

Locomotion The human colon is a long channel of varying shape and diameter, whose walls

can be silky smooth at one section or thrown into turbulent folds in another, yet at some points can be dry and rough. To make matters worse, the layout of the colon consists of unpredictable flexible 3-D curves and bends, which are near impossible to describe mathematically or to model simulations accurately. To design a robot that can accommodate such variations and propel itself through the entire organ poses great challenges. From reviewing other researchers’ work, clinical observations, and practical experience, the following criteria are proposed when building a robotic colonoscope. The body of the robot must be flexible enough to conform to the acute bends found in the colon. Any rigid distances must be kept to a minimum. Generally, the robot’s body surface (excluding the propulsion mechanism) must be smooth and well lubricated to reduce friction as it slides against the colon walls. The rigid diameter of the robot should not be greater than 29 mm, which is the smallest average internal diameter of the colon (rectal sigmoid). The robot must be capable of propelling itself right up to the caecum for a thorough colonoscopy examination. = The propulsion mechanism is preferably arranged at the distal end of the robot, so that its path will not be restricted by the curves and bends found in the colon. Any mechanism used to grip onto the colon walls must be blunt and pref-

.

.

7

d

b b

d I

I

(Pressured)

.

.

7. Basic structure of rubber actuator. (Adapted from [4].)

(a) Initial Configuration

(b) Straight Movement

(c) Bending Movement

8. Basic robot structure and movement. (Adapted from [4].) 88

IEEE ENGINEERING IN MEDICINE AND BIOLOGY

November/Detember 1997

_.

erably made of a soft material (e.g., rubber). Hard objects with sharp edges will easily damage the delicate colon walls. The robotic colonoscope must have cavities running through its length to allow optical fibers, aidwater tubing, and surgical tools to pass through to its distal end. Koji Ikuta, et al. [3], pioneers in this field, used shape memory alloys (SMAs) to develop an automated endoscope in 1988. They made use of the resistance of the SMA in a feedback control scheme to guide the snake-like robot (Fig. 5) around obstacles. The SMA springs were connected mechanically in parallel, but electrically in series. This arrangement increased the absolute value of electric resistance of the SMA, without any reduction of its performance. This also eliminated the need for sensors such as potentiometers and encoders. The basic design of the active endoscope model was done by considering its application to a fibersigmoidscope. For this purpose, the endoscope has enough mechanical compliance to pass smoothly through the sigmoid colon, which has the smallest radius of curvature. Figure 6 shows a picture of the experimental model. It has a diameter of 13 mm, which is comparable with endoscopes in the market of 10 to 20 mm. Ikuta believes that the colon is mainly two dimensional. Thus, this model is designed to have five segments, comprised of four segments with flexibility in the same direction on a plane and one segment of the tip which can bend orthogonally to this plane. The driving mechanism of each segment consists of a stainless-steel coil spring, which acts as the main skeleton at the center of a joint, and a series of SMA coil springs arranged around the joint. In this model, each segment has one degree of freedom, so that a pair of SMA actuators, which are capable of antagonistic motion, are arranged in symmetry with respect to the axis. It is this antagonistic activation of the SMA springs that brings about the required bending motion. From further experiments, Ikuta, et al., discovered that an antagonistic type resistance-feedback control produced the fol-

.. .

1.

fn 2.

\U

3.

4. 3.1

I

5.

6.

I

(a) Forward Motion Mode

(b) Bending Mode

9. (a) Forward motion mode; (b) bending mode. (Adapted from [4].) Rubber Tube--\

r Nylon Sleeve

Tip Support

I

I

Heater 1

10. Drive actuator mechanism using hydrogen sttorage alloy. (Adapted from [4].)

I /

P z i o n Aperture Spine Advance/Retract Tip Control-\ Spine Axial Travg L i m y Spine Passage Flextural Tip Cointrol

7

--I

I

\

rSninn

lowing advantages:

Reduced hysteresis of SMA actuator Lower chance of overheating Drastic improvement for robustness against changed conditions of heating and cooling Sensorfunctionfor positionalcontrol

.

November/December 1997

Endoscope Conduii

I

Proximal

I

Spine Stiffen/ilelax Control J

11. Cross-section of endoscope with controllable stiffness spine. (Adapted from [5].) IEEE ENGINEERING IN MEDICINE AND BIOLOGY

89

Aluminum Cvlinder

Flexible Cable

Ceramics Sphere

Joint j Reaction Force

12. (a) Alternating bead-shape sequential chain figure; (b) Bead chain with a continuous cable. (Adapted from [5].)

13. Photograph of a robotic endoscope. (Adapted from [6].)

14. Inserting a prototype robot into the intestine of a pig. (Adapted from 161.)

,- ODtical

the basic actuator. It is made of a rubber tube covered with a nylon sleeve and two tip supports. By varying the pressure of air inside the rubber tube, the actuator can be elongated or shortened. The elongation action is similar to that of inflating a “long” balloon. The rubber-bag actuator is capable of high displacements (stretch ratio of over 50%) and high power for its simplicity and light weight. Its high flexibility enables it to bend in any direction without being damaged. By arranging bundles of eight rubber actuators together, end to end, in three separate units an in-pipe robot can be made (Fig. 8). By a sequence of pressurizing and depressurizing the rubber actuators, the robot can be made to crawl into a pipe and turn at right angles. Figure 9(a) shows the sequence for the forwardmovement of the robot from right to left. It is similar to how an inch womi would move, a form of locomotion later used by maily other researchers. The sequence of operation is: 1. The rear unit is inflated to adhere to the pipe while the middle and front units stay depressurized. 2. The middle unit inflates to propel the robot forward. 3. The front unit inflates to adhere to the pipe wall. 4. The rear unit depressurizes to release its hold on the pipe wall. 5. The middle unit depressurizes to pull the rear part of the robot forward. 6. The rear unit IS once again inflated and the whole procedure repeats. The backward motion can also be achieved by reversingthe whole sequence. Figure 9(b) shows how the robot can be made to go around a bend, with the following sequence: 1. The rear unit inflates just before the bend, while the other units remain depressurized. 2. Only two of the eight actuators of the middle unit inflate so that the robot bends 90” to enter the vertical p ~ p e .

CCD Signals Bellow

Microcamera

~

15. Scheme of the microrobot for colonoscopy. (Adapted from [7].)

also discovered that it is essential e the heat generated by the SMA by using cooling water or air. This is to reduce the response time and also to maintain a constant low temperature. Toshio Fukuda, et al.[4], developed an in-pipe inspection robot that is capable of 90

inside pipelines of nuclear power and chemical plants. It uses rubber gas actuators that are lightweight and flexible enough to make very acute turns. Although this robot was designed for industrial applications, it could be modified for use as a colonoscope. Figure 7 shows IEEE ENGINEERING IN MEDICINE AND BIOLOGY

3. The front unit inflates to get a strong grip on the pipe wall. 4. The rear unit deflates. 5. The middle unit deflates so that the robot is pulled back into its original shape. 6. The rear unit inflates and the robot can proceed in the forward-motion mode. The control of the pressure of hydrogen gas into the rubber actuators is done with the aid of hydrogen storage alloys. This alloy can reversibly absorb or release hydrogen gas to generate the required NovembedDetember 1997

equilibrium release pressure. It can absorb up to 0. 15m3/kgof hydrogen and generate up to 0.6 MPa under normal conditions. In his model robot, Fukuda, et al., placed the alloy, lanthanum-richmishmetal, in an air-tight container that was connected to the rubber actuator, as shown in Fig. 10. Since the absorption ability of the alloy is totally dependent on its temperature, a heater based on variable current is used. The heater, along with a thermister, is connected via an A/D converter to a computer to form a closed-loop temperature-control system. Fukuda, et al., designed this robot for inspection of metal pipes in chemical or nuclear power plants. Its dimensions are suitable for use in pipes of two inches inner diameter. However, it can be seen that the movement patterns of the robot are similar to the needs for an automated colonoscope. With proper improvisation and miniaturization, an automated colonoscope could be developed similar to this pipe-inspection robot. Sturges, et al. [ 5 ] ,designed a flexible, tendon-controlled bead-chain device for endoscopy. This design employs what Sturges calls a “slide motion scheme” to traverse the device into the colon. In this scheme, the robot consists essentially of two major parts: one or two “spines” and an endoscope conduit, which is a covering tube for the spine (Fig. 11). The spine consists of a set of cylindrical beads strung together on a flexible cable. One of the most simple shapes is a cylindrical bead with a hemispherical head and an inverted conical tail, as shown in Fig. 12(a). All beads are free to rotate around their centers on adjacent beads. In the presence of a cable tension force, these beads slide axially along the cable until the positional constraints at both ends of the bead chain are satisfied. Consequently, increasing the cable tension force creates friction forces between beads and ultimately increases the apparent stiffness of the entire bead chain. In summary, pulling the cable stiffens the bead chain, and relaxing the cable tension force loosens it. Figure 12(b) shows a stiffened bead chain. The endoscope conduit is an elastomer tube with sufficient passive stiffness so that

it can be pushed forward over the stiffened spine without buckling. This outer tube is also radially stiff, so that its circular cross-section can be maintained during bending. A lubricant is supplied at the contact layer between the spine and the flexiNovember/Detember 1997

[

Front Clamping

1I

1]“

BackRelease

I

Backclamping

I

I

16. Sequence of inchworm propulsion steps of the imicrorobot. (Adapted from [7].)

SMA Spring Actuator

Open Position1

7 \

Shutter 7

Air Out

I-

I

\

\

I l l

fL

Electrical Connections

Closed Position

17. The SMA actuated microvalve. (Adapted from [7].)

ble conduit so that the relative sliding motion between them can be realized while minimizing the thrust forces. The distal end can be steered like the conventional colonoscope to allow the endoscopist to observe and point in the desired direction. In its initial position, the spine is advanced manually (or automatically) to its maximum limit (about 5 cm) and made rigid. The conduit is then inserted manually into the colon up to the first substantial turn While the spine is kept stiff, the flexible conduit moves incrementally relative to the spine, within predetermined axial travel limits, using the spine as a guide. The flexible conduit is inserted further at the same forward rate that the spine is retracted; thus, the spine is relatively IEEE ENGINEERING IN MEDICINE AND BIOLOGY

stationary with respect to the patient’s gut. The total insertion distance of the endoscope is equal to the axial spine travel limits. Such limits would be adjusted to meet the specific requirements for the radius of curvature at each bend of the colon. When the incremental forward motion of the flexible conduit is complete, the spine is relaxed and pushed forward to its maximum limit. This limit can be obtained because the conduit assumes the shape of the colon through gentle static contact. These motionless contacts form intermittent position constraints that serve as a guide for the relaxed spine. Reaction forces from the gut walls are directed toward the center line, which aids the motion of the spine inside. The spine is advanced to its maximum limit while the flexible conduit re91

mains stationary with respect to the patient. The spine is then stiffened in its new position. Advancement of the conduit and spine is now repeated cyclically; the endoscope is manually advanced no more than the spine axial travel limit with each cycle. Joel Burdick, et al. [7],invented a robotic endoscope that uses inflatable balloons and rubber bellows as actuators. It is comprised of a plurality of segments attached to each other through an articulated joint. Once again, the inchworm mode of locomotion is employed in this design. Figure 13 shows a photograph of the prototype. In order to propel itself, this robotic endoscope employs mechanisms along its length that can be described as “grippers” and “extensors.” The former are toroidal, inflatable balloons that are attached onto the outside of each segment. The primary purpose of the grippers is to provide traction against the lumen wall by expanding radially outward. Between the balloon seg-

ments are rubber bellows that act as extensors. They extend or retract like pneumatic cylinders when high- or low-pressure air is introduced, respectively. As its name implies, the extensors provide extensibility between the grippers, thus causing the robot to locally extend or retract in length. Each actuator, extensor, or gripper is controlled by its own miniature solenoid valve located within the robot itself. A control bus extends through the robot, linking all solenoid valves. This bus is connected to a controller and receiver/transmitter that controls the movement of the robot as a whole. During its operation, the robot is inserted into the colon. The robot moves forward through the lumen under its own power, using the inchworm mode of locomotion. This inchworm movement is brought about by bracing one or more segments against the lumen wall by selectively inflating some or all of the grippers. This bracing action must be strong

enough to prevent slipping of the robot. Next, some of the extensors are extended to push the deflated balloons forward. The initially inflated grippers are then deflated while the other grippers are inflated to brace the lumen, after which the extensors are retracted to pull the deflated grippers forward. This set of actions is repeated continually to bring about the forward traversing movement of the robot. Backward traversing movements can also be brought about by reversing the steps in the cycle. Depending on the number of grippers and extensors built into the robot, the sequence involved in the inchworm mode of locomotion can be extensively varied. Joel Burdick and his team managed to test a prototype robot by inserting it into the intestine of a pig, as shown in Fig. 14. Their ongoing work is focused on the design and fabrication of a new modular endoscope robot for in vivo locomotion experiments. Their short-term goal is to reduce the size of the robot as much as is practical. To achieve a reduction in size,

(A)

Back end balloon inflated to grip onto colon wall. Inner core is pushed from outside to traverse front end baloon forward.

(B)

Front end balloon moves forward.

(c)

Back end baloon deflated and front end balloon inflated.

(D) Inner core IS pulled from outside so that back end balloon is traversed forward together with the outer sheath.

(E) Back end balloon is inflated and front end balloon is deflated. The cycle repeats itself.

18. Locomotive sequence of self-propellingcolonoscope. 42

IEEE ENGINEERING IN MEDICINE AND BIOLOGY

NovembdDecember 1997

r -Fiber

Tubes

19. Cross-sectional view of alternative steerable segment. (Adapted from [9].)

they are currently developing a new generation of miniature pneumatic valves, since the size of the on-board valves represents the limiting constraint on overall endoscope size. P. Dario, et al. [7], developed a microrobotic system for colonoscopywhich, like Burdick’s robotic colonoscope, uses the inchworm mode of locomotion. This microrobot (Fig. 15) is composed of three main modules: a mothership that incorporates the propulsion system, a control module that is integratedin the mothership,and a microarm that is dedicated to the orientation of a CCD-based microcamera. The microrobot is cylindrical in shape with a diameter of 15 mm. It is 42 mm and 80 mm long in the retracted and extended configurations, respectively. Exiting from its proximal end is a tail consisting of a flexible cable housing the electrical connections for the actuators, pressurized air, and a working channel to insert microinstruments and for water flushing. The locomotion system is comprised of two modules; clamping and extension. The former clamps the robot onto the colon wall. Located at both ends of the robot, clamps are made of Plexiglas and are cylindrically shaped; 15 mm in length and diameter. The clamping mechanism is based on the suction provided by a

21. Bending movement of FMA. (Adapted from [lo].) November/Detember 1997

A X

Caps

L-. Chamber 1

20. Basic structure of FMA. (Adapted from [lo].) number of small holes disposed along the miniaturized. Currently, the team is lookactuator’s surface. The prototype clamp- ing into ways of fabricating an integrated ing actuator is comprised of four series of multivalve module to be incorporated into eight holes with diameter of 1 mm. When the mothership. a vacuum is introduced, the negative presW. Walter, et a1.[8], from the Rochessure at the small holes will cause the ter Institute of Technology, also develclamping actuator to “suck” onto the co- oped ain automated colonoscope. Similar lon wall, thus adhering the microrobot. In to Burdick‘s robotic endoscope, balloons vitro experiments determined the nega- are used in their design to grip onto the cotive pressure required to be about 8 x lo4 lon walls. To lengthen and shorten the Pa. The central module is used for exten- colonoscope, a push-pull flexible rod is sion. It is formed by an inflatable plastic used. The back-end balloon is connected bellows, which is flexible and is thus able to the outer sheath of the flexible rod, to passively bend and adapt its shape to whereas the front-end balloon is conthat of the colon. The pressure required to nectedl to the inner core of the flexible rod. inflate the bellows is 6 X lo4Pa. By a se- A pneumatic cylinder is used to drive the quence of activating the extension and core in and out of the outer sheath. By emclamping mechanisms, the microrobot ploying the inchworm method, the robot can traverse up the colon using the inch- can be propelled into the colon, as illusworm method of locomotion, as illus- trated in Fig. 18. trated in Fig. 16. Sirice the activation of the extension To regulate air flow into the microro- mechanism is from the proximal end and bot, Dario, et al., developed a pneumatic outside the patient’s body, extension microvalve actuated by an SMA mi- and retraction motions are more positive crospring (Fig. 17). This microvalve is to and more robust than most of the earbe mounted into the microrobot. The lier-mentioned designs. However, due valve is normally closed and is composed to the presence of relative motion of the of a steel microshutter that occludes the push-pull rod with respect to the colon air conduit by pressing on a seal. The walls, friction may be a concern. FricSMA microspring acts as an actuator to tional force depends on the area of conopen and close the shuttle. These compo- tact between the flexible rod and the nents are enclosed in a plastic frame for colon walls. It is also dependent on the electrical insulation.

degree of curvature through which the

The prototype valve measures 6 mm and 19 mm in diameter and length, respectively. The SMA microspring used has a diameter of 1.45 mm and can operate at an inlet gauge pressure of 0.3. Dario, et al., feel that this microvalve can be further

rod is made to bend. Excessive friction may cause the traversing mechanism to fail. Furthermore, buckling may occur at the distal end of the push-pull rod if the stroke, pushing the front-end balloon forward, is too long.

IEEE ENGINEERING IN MEDICINE AND BIOLOGY

93

22. Examples of holding modes. (Adapted from [lo].)

SMA Wire

A

r-- Diameter d Bond

/

Wire

/Lumen

Lead Wire /

/ Length L

23. Basic structure of the microactive catheter (MAC). (Adapted from [ll].)

Steerable Distal End Traversing a colonoscope from the rectum to the caecum of the colon is only part of the journey toward automation of colonoscopy. It is important for the distal end of the automated colonoscope to be able to flip or be steered toward a desired direction. This is to allow the endoscopist to inspect all portions of the colon wall more closely or to maneuver a surgical 94

tool during surgery. By turning the angulation knobs on the control device, the distal end of the conventional colonoscope can be flipped in any specified direction, as required by the endoscopist. It also has the ability of flipping more than 90°, which some endoscopists use to hook onto difficult bends as an aid in the pushing-in action. The drawback of this mechanism is that it is worked by wires IEEE ENGINEERING IN MEDICINE AND BIOLOGY

running throughout the entire length of the colonoscope. These wires require a relatively rigid case for support, thus limiting flexibility of the scope. Some researchers have proposed alternative steerable mechanisms, which can be operated autonomously. Besides developing his robotic endoscope, Joel Burdick, et al. [9], proposed, in his patent an alternative distal-end design. This design is a modification of one of his robotic endoscope’s segments, described earlier. The embodiment consists of four distinct inflatable sacs, as illustrated by Fig. 19. These sacs, which are comprised of an elastic material such as latex, are circumferentially located around a central core. This core contains a high-pressure compressed line, a low-pressure or vacuum-gas return line, and a control bus. Each sac is inflated or deflated by the action of valves. By controlling the relative pressure distribution in the sacs, the segments cannot only extend but also bend actively. Burdick, et al., propose to attach this mechanism at the distal end of his robotic endoscope. (N.B. The growing incidence of Latex sensitivity in various populations will preclude the use of this material in any device that comes in contact with a person.) Koichi Summon, et al. [lo], developed a flexible microactuator (FMA) driven by electro-pneumatics or electrohydraulics. The FMA has three degrees of freedom: pitch, yaw, and stretch. It is made in the likeness of the human finger. Thus, by serially connecting FMAs together, a miniature robot manipulator can be made. On the other hand, FMAs combined in parallel act as a multifingered rohot hand. Figure 20 shows the basic structure of an FMA actuator. There are three chambers made of fiber-reinforced rubber in the main body. The internal pressure of each chamber is independently controlled through flexible tubes that are connected to pressure control valves (pneumatic or hydraulic). Since the rubber tube is reinforced only circumferentially, the FMA deforms easily in the axial direction while it resists deformation in the radial direction. When the internal pressure in all three chambers is increased equally, the FMA stretches in the axial direction. On the other hand, when only one chamber is pressurized, the FMA will bend in the direction away from the chamber. Thus, by controlling the pressures of individual chambers, the FMA can be made to pitch, November/December 1997

24. Bending principle of the MAC. (Adapted from [ll].)

yaw, or stretch in any desired direction. Figure 21 shows the bending movement of a 1 mm diameter FMA. Figure 22 shows some uses of the FMA. Here several FMAs are combined together to act as “object holding” robots. This can be done in three ways: pinching, pair-pinching, or grasping, depending on the types of objects to be held. These different actions are brought about by individually controlling each FMA. With proper modifications, the FMA could be used as an actuator to steer the distal end of an automated colonoscope because: The structure is simple and it would be easy to miniaturize the FMA. This is especially important in medical robotics. The smaller the surgical tool, the less damage is done to human tissue. In spite of its size, the FMA has a relatively high power density. To move heavier objects, hydraulics would be used instead of pneumatics. It has a quick response time, which is required in real-time colonoscopy. Having many degrees of freedom, it is suitable for robotic mechanisms. For more intricate and difficult movement, several FMAs can be joined together in series. The FMA is inexpensive to manufacture, thus easily available to all types of industries. Unlike mechanical actuators, the FMA operates smoothly and gently because of

.

.

. .

.

November/Detember 1997

its frictionless mechanism. Being “gentle,” the FMA is good for tasks requiring human touch, such as picking up an egg. Fukuda, et al. [ 111, in their bid to improve the conventional wire-guided medic a l c a t h e t e r , have developed a microactive catheter (MAC) with two degrees of freedom. The basic structure of the MAC is shown in Fig. 23. The MAC is basically made of three strips of SMA wires embedded at 120 intervals in a cylindrical housing made of elastic material. Both ends of the SMA wires are fixed by bonding and connected to lead wires to carry an electric voltage. To prevent the leakage of electric current into the patient’s body, the tip of the MAC is waterproofed. The diameter of the MAC can be as small as 2 mm. When an electric current is introduced into one of the SMA wires, it will be heated. In doing so, it will shorten in length, causing the entire MAC to bend away from its central axis, as shown in Fig. 24. Once the electric current is shut off, the SMA wire will start to cool. Depending on the elasticity of the housing material, which acts as the biasing force, the SMA wire will return to its original length, causing the MAC to straighten again. The angle of bend, 0, is dependent on the current carried by the SMA wires. Thus, by individually controlling the flow of electricity into the three SMA wires, the MAC can be made to bend in any desired direcIEEE ENGINEERING IN MEDICINE AND BIOLOGY

tion at a specific angle, 0. In order to increase the bending angle, Fukuda, et al., proposed joining several MACs together in seriecs. Experiments have showed that with three MACs connected together in series, an angle of bend of nearly 80” is possible. However, if the concept of MAC were used to steer the distal end of the colonoscope, it must be modified to move the greater load. The transient period required for the SMA wire to reach its equilibrium position should also be as short as possible. Furthermore, there must be an efficient built-in cooling system to prevent excessive heat produced by the SMA wires, which may be a safety hazard. P. Dario, et a1.[7], proposed a novel method of designing the steerable distal end of an automated colonoscope. Their idea was to install two SMA actuated microarms at the distal end of a microrobot (see Fig. 15). One microarm maneuvers the CCD camera and the other maneuvers the surgical tool. Each microarm has two rotational degrees of freedom, which are orthogonal to the robot axis. Again, by controlling the current through the SMA actuators, these microairms can be bent in any desired direction. Experimental tests have demonstrated that these microarms are capable of bending up to 4.5’around the two axes.

Conclusion This;survey discussed the work done by researchers in the two aspects of colonoscopy automation: locomotion and steerable distal end. The results of some of the researchers seem very promising. In vitro and in vivo experimentations have been carried out to prove the possibilities of a robot moving along a patient’s colon, treating polyps as they are encountered. The authors believe that conventional colonoscopy will shortly be revolutionized.

Acknowledgments The authors would like to thank the following researchers for their permission to publish their works; Ikuta K, Fukuda T, Sturges R H, Burdick J W, Dario P, Walter W W and Suzumori K. Special thanks also goes out to Brian Davies and Francis Seow-Choenfor their valuableinputs. This work is a sub-project of smart material research funded by Singapore NSTB’s sb-a-

tegic research fund (JT ARC 7/96). So0 Jay Phee graduated from Nanyang Technological University, Singapore, with a B.Eng in Mechanical and Produc95

tion Engineering. He is currently pursuing his M.Eng at the same university, majoring in medical robotics. His candidature is sponsored by the Singapore Government. Dr. Wan Sing Ng has been a lecturer in the School of Mechanical and Production Engineering at Nanyang Technological University, Singapore, since 1993. He obtained his Ph.D. in medical robotics from Imperial College of Science, Technology, and Medicine, London, UK. In 1991, with a group led by Dr. Brian Davies, he successfully implemented an active surgical robot to perform transurethral resection of the prostate using electrosurgical means. Dr. Ng now leads a group of researchers in Nanyang Technological University researching medical robotics and computer assistance in surgery, in collaboration with the Imperial College as well as industry partners. Among the efforts are: a robot for urology using various energy means (laser, electricity, microwave) to treat benign prostate hyperplasia and prostate cancer; a robotic colonoscope; and interventive augmented reality systems for urological and orthopedic applications.

Dr. I-Ming Chen received the B.S. degree from National Taiwan University in 1986, and the M.S. and Ph. D. degrees from California Institute of Technology, Pasadena, CA, in 1989 and 1994, respectively, all in mechanical engineering. Since February 1995,he has been with the School of Mechanical and Production Engineering in Nanyang Technological University in Singapore as a lecturer. His general research interests are in modular reconfigurable robots and automation tools, medical robotics, smart material-based actuators, and robot locomotion. He is a member of IEEE and ASME. Dr. F. Seow-Choen is currently Head & Senior Consultant of the Department of

96

Colorectal Surgery, S i n g a p o r e General Hospital (SGH). He is also concurrently Director of the Endoscopy Centre at SCH. He has published more than 100 scientific papers as well as chapters in four books on various aspects of colorectal surgery. He is on the Editorial Board of the British Journal of Surgery and Techniques in Coloproctology and is the “1997 Outstanding Researcher” of SGH. Besides colorectal surgery, Seow-Choen is also well known as an avid amateur entomologist.

rehabilitation robots. Dr. Davies is a member of the U.K. Engineering and Physical Sciences Research Council’s Medical Engineering Collegiate and is Vice-Chair of the UK Institution of Mechanical Engineers “Engineering in Medicine” Division. Address for Correspondence: WanSing Ng, Nanyang Technological University, School of Mechanical and Production Engineering, Systems Engineering Division, Nanyang Avenue, Singapore 639798, Republic of Singapore.

References

Dr. Brian Davies is a Reader in Robotics in the Department of Mechanical Engineering at Imperial College of Science, Technology and Medicine, London, UK. He has a Ph.D. in Medical Robotics. He has been at Imperial College since 1984, prior to which he was a lecturer at University College, London, where he researched powered prosthetics and rehabilitation robotics. At Imperial College, his research interests have included a number of “industrial robots” for manufacturing apphcations. Research into rehabilitation robots for the disabled has continued, but the primary medical robot research activity is now in robotic and computer assisted surgery. He heads the “Mechatronics in Medicine” group, which researches a wide range of mechatronics devices applied to surgery. A specially developed robot for prostatectomies was applied clinically in 1991 (a world first in removing quantities of tissue from human patients). The system has been recently extensively modified and clinical trials successfully completed at Guys Hospital, London. Another project is concerned

1 Silverstein E and Tytgat NJ: Atlas of Gastrointestinal Endoscopy 1987 2 Church JM: Endoscopy of the Colon, Rectum, andAnus 1995 3 Ikuta K, Tsukamoto M and Birose S: Shape memory alloy servo actuator system with electric resistance feedback and applicahon for achve endoscope Proceedings of the 1988 IEEE International Conference on Robotics and Automation, pp 427-430, 1988 4 Fukuda T, Hosokai H and Uemura M: Rubber gas actuator driven by hydrogen storage alloy for in-pipe inspection mobile robot with flexible structure Proceedings ofthe 1989 IEEE International Conference on Robotics arid Automation, pp 1847.1852, 1989 5 Sturges RH and Laowattana S: A flexible, tendon-controlled device for endoscopy Proceedings of the 1991 IEEEInternational Conference o n R o b o t i c s and A u t o m a t i o n , p p 2582-2591, 1991 6 Grundfest SW, Burdick JW and Slatkin AB: The development of a robotic endoscope Proceedings of the 1995 IEEE International Conference on Robotics and Automation, pp 162-171,1995 7. Dario P, Carrozza MC, Lencioni L and Magnani B: A microrobot for colonoscopy Proceedings of the Seventh International Symposium on Micro Machine and Human Science, pp. 223-228, 1996 8 Walter WW, Lakshman V and Amon JN: S elf -pr o p el l i n g c olono s c op e h t tp // www rit edu/jnantl/asmeabs html , 1995 9 Grundfest SW, Burdick JW and Slatkin AB: Robotic endoscope United States Patent, Patent

with a special purpose robot for knee sur-

Number 5,337,732, 1994

gery, to accurately cut the bones of the knee for prosthetic implants. Other projects have been concerned with a feasibility study into robotic neurosurgery; a low-cost targeting system for kidney stone removal; and a fundamental study of telemanipulator penetration in soft tissue surgery. Dr. Davies has published extensively on the above topics, as well as on safety issues for robot surgery and on

10 Suzumori K, Iikura S and Tanaka H: Development of flexible microactuator and its appli cations to robotics mechanisms Proceedings of the 1991 IEEE International Conference on Robotics and Automation, pp 1622.1621, 1991 11 Fukuda T, Guo S, Kosuge K, Arai F, Negoro, M and Nakabayashi K: Micro active catheter system with multi degrees of freedom Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pp 22902295, 1994

IEEE ENGINEERING I N MEDICINE AN0 BIOLOGY

November/December 1997