Development of a novel endoscopic manipulation system: The Endoscopic operation robot

Abstract

AIM: To develop and evaluate the endoscopic operation robot (EOR). The EOR is a robot system designed specifically for remote manipulation of the scope during gastrointestinal endoscopy by a seated endoscopist.

METHODS: Total colonoscopy examinations using a colonoscopy training model were performed compared conventional insertion by manual manipulation and remote-controlled insertion, using the EOR. The author investigated the time taken for each of the 50 examinations.

RESULTS: The median insertion time (in minutes) for each 10 examinations (EOR vs manual manipulation) was 73.70 ± 25.37 vs 3.77 ± 1.34 in the first group, 38.40 ± 6.24 vs 3.40 ± 0.97 in the second group, 27.6 ± 4.01 vs 2.70 ± 0.95 in the third group, 23.8 ± 3.65 vs 3.10 ± 0.88 in the fourth group, and 22.9 ± 5.02 vs 2.60 ± 1.08 in the fifth group.

CONCLUSION: The study suggested the possibility of the clinical application of the EOR.

Key Words: Hepatitis B; Hepatocellular cancer; Hepatitis B surveillance; Vaccination; Screening

INTRODUCTION

With an ever-expanding range of indications requiring minimally invasive therapy in the form of therapeutic gastrointestinal endoscopy, endoscopic targets and procedures are becoming more complex and broad ranging. Consequently, the required level of endoscopic precision is rising, and the duration of endoscopy procedures is lengthening. Endoscopic submucosal dissection (ESD)[1,2], natural orifice transluminal endoscopic surgery (NOTES)[3], and other gastrointestinal endoscopic techniques for minimally invasive therapy, lighten the burden on the patient, but increase the burden on the endoscopist in terms of expertise, dexterity, and proficiency. Many ways to reduce the burden on the endoscopic surgeon through the use of telesurgical units, such as da Vinci, developed by Intuitive Surgical, and Zeus, developed by Computer Motion, and other well-known endoscopic robots[4,5], have been developed. These robots, however, are specifically designed for surgeons using rigid endoscopes. In contrast, there have been no reports on the development of true robots which have been specifically designed for the flexible endoscopes which are required for oral or anal approaches in gastrointestinal endoscopy, other than a robot specifically designed for NOTES, primarily for forceps manipulation[6-8]. Research and development related to gastrointestinal endoscopic therapy has generally focused on tools attached to the endoscope, and surgical tools inserted in the channels[9]. The author has developed and reported several such tools, including an irrigation hood[10-12], an endoscopic aspiration mucosectomy (EAM) hood[13-18], and various ESD devices[19-26].

Further to this, the author has developed a new endoscopic operation robot (EOR) for full robotic manipulation of every procedural element of gastrointestinal endoscopy, including all the basic elements as well, thus eliminating the need for direct physical contact with the endoscope (Figure 1).

Figure 1 The system of the endoscopic operation robot and colonoscope training model (f). It consists of the main unit (a), the manipulation unit (b), the aspiration control unit (c), the light source unit (d) and the aspiration unit (e)

In manual endoscopy, the grip of the endoscope is held in the left hand and the vertical angulation (up-down) knob and the horizontal (right-left) angulation knob are manipulated with the fingers of the left hand, thus curving the endoscope tip vertically and horizontally. The tip is rotated by oscillation of the left wrist, and tip extension and retraction are performed by horizontal manipulation in the long-axis direction using the right arm while gripping the insertion unit. It is thus a “four-axis” manipulation, performed by the endoscopist while standing. The technique used for this manipulation input varies with the endoscopist, in terms of individual postures, habits, and customary practices, but these differences cancel out in the gastrointestinal tract, where they are output as mechanical movement. In short, manipulation input tends to vary with the individual endoscopist, but the variations mutually cancel in the output, to obtain relatively simple endoscope movements in the four axial directions. However, these individual differences tend to complicate the necessary acquisition of multifaceted skills by the endoscopist.

The EOR was developed to further the mechanization of this manipulation, with the aim of simplifying the operation by the endoscopist, reducing or even eliminating individual differences, and to facilitate the standardization of endoscope manipulation. This report describes the EOR concept and configuration, as well as its evaluation in complete colonoscopy examinations using a colonoscopy training model.

MATERIALS AND METHODS

System configuration

The EOR consists of the main unit (Figure 2A and 2B), the manipulation unit (Figure 2C), and the aspiration control unit (Figure 2D), all three of which are newly developed, together with (Figure 2D) the light source unit and (Figure 2D) the aspiration unit, both of which are pre-existing devices. The manipulation unit includes a monitor, two joysticks, and three foot switches. The right joystick controls the up-down and right-left angulation knobs, and the left joystick controls tip rotation, extension, and retraction. The three foot switches control the air supply, air suction, and water supply. If the endoscopist’s hands are removed from the joysticks, the endoscope simply remains in position, without automatically returning to a neutral position.

Figure 2 The system of the endoscopic operation robot. A: The left part of the main unit of the endoscopic operation robot (EOR) has four motors; the first motor controls up-down angulation (a), the second motor controls right-left angulation (b), the third motor controls rotation (c) and the fourth motor controls extension and retraction (d); B: The right part of the main unit of the EOR is the insertion part of the endoscope; C: The manipulation unit of the EOR. It includes a monitor, two joysticks, and three foot switches (no photos). The joystick on the right controls the up-down and right-left angulation knobs, and the joystick on the left controls tip rotation, extension, and retraction; D: The aspiration control unit (a), the light source unit (b) and the aspiration unit (d); E: Colonoscope training model produced by KYOTO KAGAKU Co., LTD. (Kyoto, Japan). EOR: Endoscopic operation robot.

The four-axis movement of the endoscope is driven by the four motors of the main unit, each via a separate timing belt and pulley transmission, thus serving the up-down and right-left angulation knobs, the rotational oscillation component, and the extension-retraction component. The endoscope is an Olympus GIF-Q230 (Tokyo, Japan), mounted on the rotational-oscillation component of the main unit. In accordance with the properties of the GIF-Q230, tip curvature control by vertical and horizontal movement of the right-hand joystick enables an up-down angulation knob range of 210° up and 90° down and a left-right angulation knob range of 100°. Rotational oscillation control by vertical and horizontal movement of the lefthand joystick enables 150° rotation of the endoscope with an effective length of 1030 mm. The power for these four-axis manipulations is provided by the four motors actuated by a specifically designed computer program.

The air supply and air suction button on the endoscope is set to ON, and the two interim valves of the aspiration control unit are connected to the suction unit and the water supply tank for the light source unit, to enable input of the air supply, air suction, and water supply via the three foot switches.

With the EOR, the endoscopist controls the operation with the two joysticks and the three foot switches in a seated position while watching the monitor on the manipulation unit, without touching the endoscope at all once the procedure begins.

Procedures: Insertion in colonoscope training model

The colonoscope training model produced by KYOTO KAGAKU Co., LTD. (Kyoto, Japan) was used (Figure 2E). This model has six training patterns (beginner’s grade 1-3, intermediate grade 1-2, and higher grade). For this study, beginner’s grade 1 was used. The aims with beginner’s grade 1 are as follows: 1) learn how to insert the colonoscope deeply into the transverse colon and the ascending colon, without forming a loop at the sigmoid colon; 2) acquire basic insertion skills required to pass through each part of the colon; 3) learn the “hooking the fold” method to pass through the sigmoid colon; and 4) learn “withdrawal” manipulation to go through the hepatic flexure.

All cases of total colonoscopic examination were performed by the author, who has completed 5000 total colonoscopic examinations.

The author investigated the records of 100 total colonoscopic examinations and compared 50 conventional insertions by manual manipulation and 50 remote-controlled insertions using the EOR. The learning curves of endoscopists using the EOR were also investigated. Learning curves were assessed as the insertion time for each 10 examinations. Insertion time was measured from the model anal region to the cecum.

The tip of the EOR endoscope was manually inserted 3 cm into the model anal region, and the endoscope was thereafter remotely controlled by the operator using the manipulator unit.

The EOR was designed by the author, and was produced by Takeshi KUROKI and Takahiro SUGIHARA at Kyushu Polytechnic Collage.

Statistical analysis

The results, insertion time for each 10 examinations, were presented as mean ± SD. An analysis of variance (ANOVA) was used to compare insertion time for each 10 examinations. Qualitative data were analyzed by the Mann-Whitney U test with Bonferroni correction. A P value of less than 0.05 was considered significant.

RESULTS

The overall complete insertion rate was 100% (100/100; 50/50 conventional insertions by manual manipulation and 50/50 remote-controlled insertions using the EOR). The median insertion time was 3.6 ± 1.96 min by manual manipulation and 37.28 ± 22.47 min by remote-controlled insertion using the EOR. The median insertion time by EOR insertion for each 10 examinations was 73.70 ± 25.37 in the first group, 38.40 ± 6.24 in the second group, 27.6 ± 4.01 in the third group, 23.8 ± 3.65 in the fourth group and 22.9 ± 5.02 in the fifth group. The median insertion time by manual manipulation 3.70 ± 1.34 in the first group, 3.40 ± 0.97 in the second group, 2.70 ± 0.95 in the third group, 3.10 ± 0.88 in the fourth group and 2.60 ± 1.08 in the fifth group.

Concerning the EOR learning curve, median insertion time was significantly shorter with each succeeding group until the third group of 10 cases, and was less than 30 min after the third group of 10 cases (Figure 3).

Figure 3 Learning curve assessed based on insertion time. EOR: Endoscopic operation robot.

DISCUSSION

In planning, designing, and commissioning the construction of the EOR, two questions that were considered and must ultimately be resolved are endoscopist familiarization and endoscopy standardization. Remote manipulation by joysticks while seated is conceptually quite different from the conventional direct manual manipulation of the endoscope while standing, and it is unlikely that an endoscopist well practiced in the manual procedure would find it easy to adapt to the EOR concept. However, many endoscopists are undoubtedly familiar with the control panels and joystick operations of video games and other such devices, and this may ameliorate some initial awkwardness, speed of learning, and heighten proficiency. In regard to the standardization of endoscopic therapy techniques, further investigation on the potential of the EOR for contribution to this goal will be necessary, but the expectation is that a robotic manipulation system, such as the EOR, will greatly facilitate general standardization of endoscopic techniques by reducing the complexities associated with direct manual manipulation of the endoscope arising from the differences among endoscopists in manipulation customs, practices, and levels of dexterity. Moreover, such a system will substantially broaden the range of applications for endoscopic therapies.

The EOR has been developed primarily for utilization in ESD, NOTES, and other orifice-insertion procedures in minimally invasive therapy. In the present study, however, colonoscopy was considered the most appropriate therapy for the initial evaluation of the EOR manipulation capabilities, due to the requirement for maximum precision in 4-axis manipulation.

The learning curve for EOR manipulation in the colonoscopy model was determined from the insertion times in the series of EOR trials performed by the author, who had had no previous experience with EOR manipulation, but who, in clinical practice, has had extensive experience in conventional manual insertion. Insertion time was used as an indication of proficiency in EOR insertion. A learning curve for manual insertion was not determined, due to the author’s extensive experience. The learning curve for EOR increased over the first 30 insertions but remained flat thereafter, giving no indication of the prospect for a further shortening in insertion time.

The primary reason for the apparently lower limit in the reduction in EOR insertion time found in these trials may be attributable to the lack of function for presentation of force and tactile sensation by the EOR in its present version. In the intestinal tract shortening maneuver, which is performed to increase insertion efficiency, reliance is placed in part on the tactile sensation of catching the intestinal tract on the curved scope tip. With the present EOR, however, this maneuver is impracticable, due to the absence of tactile sensation. In the absence of feedback-induced control in a clinical setting, an unintended application of force could increase the risk of pain and possibly perforation. It will therefore be necessary to consider the incorporation of kinesthetic and haptic feedback presentation functions into the EOR, together with control systems providing a slight degree of play in the joystick and a target tracking or other function providing automated supplemental control of endoscope tip movement.

The EOR nevertheless has the potential for achieving modes of manipulation that cannot be achieved by manual manipulation of conventional scopes, along with other functional advantages. With the continuing advances in endoscopic therapy, the length, complexity, and proficiency of the related procedures are testing the limits of endoscopists with regard to maintaining their field of vision, and the skill required to coordinate the manipulation of single general-purpose endoscopes. Through the integration of all scope and device manipulations in a single control console, along with breaking down the coordinated manipulations, and allowing seated operation, the EOR holds the promise of substantially reducing the burden on the endoscopist. The breakdown of coordinated manipulations refers to capabilities such as being able to fix the field of vision by the operator, having removedhis or her hand from the joystick, and, as circumstances require, the capability to limit manipulation specifically to the treatment tools.

The advantage of the EOR relating to maintenance of the insertion axis was clearly demonstrated. In conventional manual insertion, maintaining the insertion axis requires manipulation of the handle by finger action and rotation by wrist action in a physiologically constrained environment, and, in some cases, the physiological limits may prevent successful insertion into deep regions. In this case, it is difficult to continue conventional insertion whilst seated. With the EOR, in contrast, the endoscope position does not change when the hand is removed from the joystick. The axis is thus maintained, and insertion to deeper regions can be readily resumed from that angle, without concern for a departure from the axis. The freedom from both the physiological constraints on the range of motion in the joints of the endoscopist and the consequent need to maintain difficult bodily postures is in fact an important advantage, particularly in therapeutic endoscopy, with the related need for manipulation of surgical tools.

Adoption of EOR-based systems for colonoscopic examination could open the way to many new modes of application. It would facilitate the development of advanced systems for EOR training on colonoscopy models, by incorporating systems for time measurement in conjunction with optical sensors appropriately positioned in the intestinal tract model for calculation of intestinal internal observation ratios in the circuit, for counting and recording incidents of simulated pain due to excessive intrusion into the model mesentery, together with a scoring system for each element of the procedure. In clinical implementations, the incorporation of insertion time measurement and input systems responsive to vital changes and the experience of pain signaled by the patient using appropriate buttons could facilitate objective evaluation of insertions and hospital performance. Ultimately, and with the provision that every aspect of safety be considered and assured, it may be possible to achieve completely automated colonoscope insertion for difficult cases, as well as for more routine cases, through the incorporation of balloon, image recognition, and other necessary sensors on the scope tip and effective computerized system control.

Other envisioned developments ultimately include the automation of ESD, NOTES, and other endoscopic therapies. However, many challenges would have to be met for these purposes. The requirements for fully automated ESD, for example, would include lesion recognition, determination of resection and peripheral incision extent, always-on recognition of appropriate resection surfaces, dissection of deep submucosal layers at specific depths, and an effective response to bleeding, breathing changes, peristalsis, and other events.

With these numerous and complex requirements, fully automated procedures remain a long-range goal. However, advances and improvements in individual component systems and devices may hasten progress. The wiper-knife was developed by the author, primarily to simplify endoscope manipulation, but it now appears highly appropriate for the EOR. The multiDOF forceps being developed for NOTES will probably facilitate many aspects of remote manipulation. With effective cooperation between medicine and engineering, it will be possible to incorporate functions such as kinesthetic and haptic feedback, presentation, target tracking, and 3D spatial presentation. With appropriate methods for adopting advances in engineering, higher levels of precision control and automation may be possible. Robotization of endoscopic manipulation such as that of EOR thus facilitates the conceptualization of endoscopic automation. At present, however, the task at hand is the continuation of research and development directed toward the identification of those component processes appropriate for automation by computerized control, and those that are appropriate for remote manipulation by the endoscopist, and their realization for the simplification of endoscopic techniques and the enhancement of their safety.

In conclusion, the EOR is a robot system specifically designed for remote manipulation in oral digestive tract endoscopy by a seated endoscopist, without directly touching the scope. Its operation, which is reminiscent of operating video-game controllers and other such devices, eliminates the physiological constraints that apply in the conventional standing-position necessary for manual endoscopic manipulation, due to the naturally limited range of motion of body joints, and it reduces the tendency for differences to arise among operators in their customary techniques and practices of endoscope manipulation. The EOR is a next-generation endoscope that is expected to bring fundamental changes to endoscopic manipulation techniques, and may ultimately lead to their automation.