bimanual coordination in natural orifice transluminal endoscopic surgery: comparing the conventional...

ORIGINAL ARTICLE: Experimental Endoscopy

Bimanual coordination in natural orifice transluminal endoscopicsurgery: comparing the conventional dual-channel endoscope, theR-Scope, and a novel direct-drive system

Georg O. Spaun, MD, Bin Zheng, MD, PhD, Danny V. Martinec, BS, Maria A. Cassera, BS,Christy M. Dunst, MD, Lee L. Swanstrom, MD

Portland, Oregon, USA

Background: The devices used for natural orifice transluminal endoscopic surgery procedures are endoscopesor inspired by endoscopic design, which makes it difficult to accomplish bimanual coordination.

Objective: We evaluated 3 operating systems in simulated natural orifice transluminal endoscopic surgery pro-cedures requiring complex bimanual coordination.

Design: Operators were required to perform an identical bimanual task by using 3 operating systems: a dual-channel endoscope (DCE); the R-Scope, which has 2 elevators for independent movement of endoscopic instru-ments; and the Direct Drive Endoscopic System (DDES), which allows separation of instruments and vision, em-ulating more of a laparoscopic surgery paradigm.

Setting: A bench-top simulation was used. Twelve teams were recruited for DCE and R-Scope testing. Twelveindividuals participated in the DDES setup. The task included 3 steps: picking up a ring, passing it between en-doscopic instruments, and placing it on a designated location.

Main Outcome Measurements: Task performance was evaluated by movement speed and accuracy.

Results: Task performance was significantly faster when using the DDES system (29 � 28 seconds) comparedwith the other operating systems (DCE: 140 � 55 seconds, R-Scope: 160 � 71 seconds; P! .001). The differencebetween the DCE and the R-Scope was not significant (P Z .370).

Conclusion: Designs that separate vision and motion have more degrees of freedom at the tip of the instru-ments, and an ergonomic user interface provides benefits for bimanual performance compared with more tra-ditional endoscopic designs. With the DDES, a single operator can perform complex endoscopic tasks fasterthan 2 operators with a DCE or R-Scope. (Gastrointest Endosc 2009;69:e39-e45.)

In the American Society for Gastrointestinal Endos-copy/Society of American Gastrointestinal and EndoscopicSurgeons White Paper1 (October 2005), the development

Abbreviations: DCE, dual-channel endoscope; DDES, Direct Drive

Endoscopic System; NOTES, natural orifice transluminal endoscopic

surgery.

DISCLOSURE: The following author disclosed financial relationships

relevant to this publication: L. L. Swanstrom serves as a consultant

for Olympus and Boston Scientific and received research support

from both companies. Supported in part by a 2007 research grant to

L. L. Swanstrom from the National Orifice Surgery Consortium for

Assessment and Research. All other authors disclosed no financial

relationships relevant to this publication.

Copyright ª 2009 by the American Society for Gastrointestinal Endoscopy

0016-5107/$36.00

doi:10.1016/j.gie.2008.12.239

www.giejournal.org

of a multitasking platform was considered a fundamentalchallenge to the safe introduction of natural orifice trans-luminal endoscopic surgery (NOTES). For most of thepublished laboratory or human NOTES cases, endoscop-ists and surgeons performed surgical procedures using de-vices that were primarily designed to work endoluminallyand followed endoscopic design.2-5 These procedurescertainly would not have been as swift and reliable asthe laparoscopic standard if performed only translumi-nally.6 Additional laparoscopic trocars2,4,5 or second endo-scopes7 were used in all reported human NOTESprocedures for both patient safety and efficiency of theoperation.

When using a regular flexible endoscope to performa complex surgical task, it is difficult, if not impossible,to obtain triangulation, to perform bimanual tasks, or to

Volume 69, No. 6 : 2009 GASTROINTESTINAL ENDOSCOPY e39

NOTES bimanual performance Spaun et al

maintain consistent imaging.8 Bimanual manipulation isessential to the success of complex surgery because it per-mits traction/countertraction, precise and efficient dissec-tion, and tissue approximation.9 When using a dual-channel endoscope, a single operator has difficulties per-forming efficient bimanual manipulation to accomplisha surgical procedure. Currently for NOTES procedures,many operators surround the endoscope trying to achievesuch bimanual coordination. This results in a severelynonergonomic user interface.4

To allow bimanual coordination, engineers are design-ing better operating platforms for NOTES. One design(R-Scope; Olympus, Tokyo, Japan), adds 2 elevators tothe end of the working channels that permit the flexiblesurgical instruments to articulate to a certain degree.This allows 2 instruments to converge on a commonsurgical target and permits some degree of traction/countertension.

Another solution is to develop a totally novel operatingsystem that not only permits independent motion of in-struments, but also separates vision and motion and givesthe operator a stable and ergonomic interface similar tothat of laparoscopic configurations. Such a system wouldallow the endoscopist to control 2 instruments and theendoscope independently, allowing bimanual coordina-tion of instruments without interfering with the view ofthe surgical site. We have had access to one such design(Direct Drive Endoscopic System [DDES], Boston Scien-tific Inc., Natick, Mass) and have conducted several inves-tigations looking at its possibilities as an endoscopicplatform.

This study was a comparative investigation of the rela-tive benefits of using the R-Scope and the DDES com-pared with a standard DCE for complex bimanualsurgical tasks in a simulation model for NOTES.

We hypothesized that the separation of vision and mo-tion, more degrees of freedom at the tip of the instru-ments, and an ergonomic user interface would supportbimanual performance better in complex NOTES tasksmeasured in a bench-top simulation.

MATERIAL AND METHODS

ApparatusThis study was conducted by using a simulation specif-

ically designed for NOTES training. This simulation re-quires surgeons to manipulate an endoscope that passesthrough a training box (mimicking the upper trunk ofthe human body) measuring 36 � 28 � 15 cm, before ma-nipulating a target. For the current study, the target wasa Sea Spikes Pod (Chamberlain Group, Great Barrington,Mass). The Sea Spikes Pod consists of flexible rubberspikes of different lengths, shapes, and colors (Fig. 1B).The endoscopes are passed through the box via structuralholes, which provide some stability against rolling but do

e40 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 6 : 2009

Capsule Summary

What is already known on this topic

The devices used for natural orifice transluminalendoscopic surgery (NOTES) procedures are endoscopes ormimic endoscopic design, making bimanual coordinationdifficult.

What this study adds to our knowledge

In a bench-top simulation comparing 3 operating systemsin simulated NOTES procedures, the best results wereachieved when the device allowed independent instrumentmovement, separation of vision and instruments, anda stable ergonomic interface.

not prevent it, similar to situations found in therapeuticendoscopy.

Task and procedureThe task included grasping a rubber ring that had been

placed on a spike with the right hand grasper (phase 1).The ring was then passed to the left grasper (phase 2),and then placed securely on another spike (phase 3).Dropping the ring by accident or on purpose to bypassthe bimanual coordination was penalized. If this hap-pened, the ring had to be picked up and placed on a spikeby using the left grasper.

The order of testing the 3 operating systems was deter-mined by the availability of the endoscopes in our labora-tory. Each participant was required to perform a bimanualtask using the DDES, the DCE, and the R-Scope. Each par-ticipant had the opportunity to practice the task once be-fore measurements were recorded.

SubjectsParticipants for the endoscope testing were recruited

from within the Legacy Health System. Team membersconsisted of a wide range of endoscopic training levels in-cluding junior and senior residents, general and laparo-scopic fellows, and attending surgeons with extensiveendoscopy experience. We also included clinical researchassistants and engineers as a novice population. A pretestquestionnaire was given to all participants to obtain demo-graphic data as well as the participants’ endoscopic expe-rience, level of training, and surgical experience. Weseparated participants into expert and novice groups.The expert group consisted of surgeons with more than7 years of endoscopic and laparoscopic experience. Thenovice group consisted of surgeons with less than 7 yearsof endoscopic experience and students with experience inNOTES labs. (We initially had only the students in the nov-ice group and the surgeons in an intermediate group.However, the students, who are well trained in endo-scopic and laparoscopic bench-top testing, performed

www.giejournal.org

Spaun et al NOTES bimanual performance

slightly better than the intermediate group of surgeons, sowe decided to pool these 2 groups together into the nov-ice group.) For the DDES testing, we also included 3 engi-neers because the DDES was new to all other participants.

When using the DCE and the R-Scope, participantswere paired into dyad teams, one to control the endo-scope and the other to run the instruments. A total of12 teams were recruited for the DCE and for the R-Scopetasks. The performance was averaged over 2 individuals’performances on a team. Because of its design as a ta-ble-mounted platform, the DDES required only 1 opera-tor. We recruited 12 participants for the DDES task.Because only 1 operator was required to operate theDDES, each participant performed the task twice, andthe performance was averaged over 2 trials.

Overall, 24 tasks were performed with every operatingdevice. The expert group performed 10 tasks usingthe DCE, 10 tasks using the R-Scope, and 8 tasks using theDDES. The novice group performed 14 tasks usingthe DCE, 14 tasks using the R-Scope, and 16 tasks using theDDES.

Figure 1. (A) Dual-channel endoscope (GIF-2T 160) placing a ring at the

Sea Spikes Pod (B).

www.giejournal.org

Devices testedWe used the nonproduction R-Scope (Olympus)

(Fig. 2) together with 2 standard Jumbo Biopsy Forcepswith Serrated Jaws (Catalog No. 1337; Boston Scientific).The diameter of the endoscope is 13.5 mm. Olympus de-veloped the R-Scope specifically for complex endoscopicprocedures such as endoscopic submucosal dissection.Two elevators were added to the 2 working channels.The resulting 1 extra degree of freedom for each instru-ment was expected to increase the ability to perform com-plex maneuvers. We used the most recent prototype(XGIF-2TQ 160R) for our task. There are 2 versions ofthe R-Scope prototype. The first version was the XGIF-2TQ 240R. This endoscope was similar to the 160R thatwe used, but it had an additional second bending sectionbehind the standard bending section to allow greater flex-ibility in tip position. It was found that this feature was notthat useful, so the 160R was developed without this fea-ture. This allowed for a thinner insertion tube, eliminatingthe extra set of tip control knobs and making the controlsection lighter and more compact. Both of these instru-ments have been used in Japan by gastroenterologistsfor performing endoscopic submucosal dissection. Theyoffer some benefit, but are not widely used for thatpurpose.

As our comparative standard, we tested a standarddual-channel endoscope (GIF-2T 160, Olympus) (Fig. 1)armed again with 2 standard Jumbo Biopsy Forceps withSerrated Jaw (Boston Scientific). The diameter of this de-vice is 13.2 mm.

The DDES (Boston Scientific) (Fig. 3) was tested withits own prototype tripod and 2 prototype flexible endo-scopic graspers and provided a 5-mm upper endoscope(N 180; Olympus) for visualization. The DDES consistsof 3 main elements: (1) a steerable flexible guide tubewith 3 channels, a 6-mm visualization channel, and 2independent 4-mm instrument channels; (2) a set of artic-ulating 4-mm instruments; and (3) an ergonomic table-mounted rail platform.

When in place, the distal portion of the DDES guidetube moves in a manner similar to that of an endoscopewith up, down, and left-right control. When an endoscopeis placed via the visualization channel, it slides, rolls, andtilts without disturbing the instruments. The proximal por-tion of the DDES guide tube is fixed on a stable rail plat-form. The instruments have control handles that glide onindividual rails. The control handles transmit hand motionto the instrument tips with 5 degrees of freedom. Withthese designs, the DDES allows independent instrumentmanipulation without disturbing the visual perspective.The outer diameter of the device is 18 � 22 mm.

Video recording, analysis, and statisticsThe endoscopes were connected to a CV-140 Processor

(Olympus) and illuminated by a CLV-V20 light source



(Olympus). The task was displayed on a monitor (SonyPVM-1953MD, Tokyo, Japan). All tasks were videotapedwith a VHS video recorder (Panasonic, Matsushita ElectricIndustrial Co., Kadoma, Osaka, Japan), and the tapes werereviewed by 2 independent reviewers.

Performance was assessed by measuring movementspeed and accuracy. Movement time started from the mo-ment when a grasper was displayed on the video monitor.Time was stopped when the ring was placed securely onthe target spike using the left-hand instrument. Whenthe ring was dropped, a time penalty of 60 seconds wasadded to the task time. For each trial, the task was brokendown into 3 phases, as defined in the previous section:reaching and grasping, transferring between hands, andplacing on the target spike.

Figure 2. Olympus R-Scope (XGIF-2TQ 160R).


To test the interrater reliability, the 2 reviewers exam-ined the same videotapes of 10 trials independently, re-viewing a total of 30 phases in these 10 trials. Two videoreviewers achieved satisfactory agreement on definingthe lengths of these phases. Twenty phases were thesame (66.7%), 8 phases varied by 1 second (26.7%), 1 var-ied by 2 seconds (3%), and 1 varied by 3 seconds (3%).The reason that the 2 video reviewers had more thana 2-second variation in deciding the movement phaseswas because they had difficulty in judging the exact timeat which the grasper let go of the ring. Cronbach’sa was .99, a high interrater reliability when calculatingthe duration of movement phases computed by 2 re-viewers. There were no differences between the reviewersconcerning the number of rings dropped during the trials.

The total time and the time for each phase were com-pared among the 3 operating systems by using a singleanalysis of variance model. A 2 � 2 analysis of variancewas used to test the impact of surgical experience ontask performance of the 3 operating systems.

RESULTS

The overall task performance with the 3 operating sys-tems is summarized in Table 1. Bimanual coordination wassignificantly faster when using the DDES system (29 � 28seconds) than the other 2 operating systems (DCE: 140 �55 seconds, R-Scope: 160 � 71 seconds; P! .001) (Fig. 4).

The difference between the DCE and the R-Scope wasnot significant (P Z .370) (Fig. 4). The DCE was signifi-cantly faster than the R-Scope in phase 2 in which the ob-ject (a ring) is passing between instruments (DCE: 57 � 30seconds, R-Scope: 91 � 50 seconds; P Z .028) (Table 1).In terms of movement accuracy, we recorded 1 ringdrop when using the R-Scope and DDES. However,when using the DCE, we recorded 5 drops. This would im-ply that the DCE is not sufficiently designed for accurateand efficient bimanual coordination.

Figure 3. Direct Drive Endoscopic System (Boston Scientific).

www.giejournal.org

www.giejournal.or


TABLE 1. Overall task performance with the 3 operating systems

P value

No. Mean (s) SD R-Scope-DCE R-Scope-DDES DCE-DDES

Phase 1

R-Scope 12 19.0 11.6

DCE 12 16.8 11.8 .595 .007 .021

DDES 12 5.3 2.5

Phase 2

R-Scope 12 90.7 49.9

DCE 12 56.9 30.0 .028 !.001 .007

DDES 12 7.4 3.4

Phase 3

R-Scope 12 48.2 19.4

DCE 12 53.5 26.0 .527 !.001 .006

DDES 12 13.9 22.4

Time

R-Scope 12 157.8 65.0

DCE 12 127.1 45.7 .139 !.001 !.001

DDES 12 26.5 22.5

Score

R-Scope 12 160.3 70.7

DCE 12 139.6 54.9 .370 !.001 !.001

DDES 12 29 27.5

SD, Standard deviation; DCE, dual-channel endoscope; DDES, Direct Drive Endoscopic System.

Differences between the expert group and thenovice group

We found that experts performed better with shorterperformance times (79 � 56 seconds) than novices (129� 108 seconds; P Z .001), averaged over all 3 groups.In detail, the expert group performed significantly betterwhen using the DCE (expert: 82 � 42 seconds vs novice:180 � 82 seconds; P Z .002). When using the R-Scope, theperformance of the expert group was better but was notfound to be significant (expert: 121 � 56 seconds vs nov-ice: 189 � 110 seconds; P Z .086). When using the DDES,the difference between the expert group and the novicegroup was not significant (expert: 25 � 12 seconds vs nov-ice: 31 � 33 seconds; P Z .583) (Fig. 5).

DISCUSSION

NOTES is an exciting concept that has the potential torevolutionize therapeutic endoscopy and minimally inva-

g

sive surgery. Early NOTES pioneers10-13 showed what ispossible when using regular flexible endoscopes. Mostof the more than 150 human cases today7 were performedwith regular DCEs and multiple operators. Most practi-tioners find it very difficult to perform bimanual coordina-tion with this type of endoscope. Bimanual coordination isdefined as the synergistic movement of 2 different instru-ments. Complex surgery typically requires bimanual coor-dination for safe completion of more technical surgicaltasks. Retraction and exposure, traction and countertrac-tion, and separation of the imaging and instrument mo-tion are basic principles of safe and efficient surgery.Our study identified 3 factors inherent in the standardDCE that contribute to the difficulty in complex taskcompletion.

First, a single operator lacks enough hands necessaryto control 2 instruments while maintaining the endoscopein the optimal position. In this study, subjects performingbimanual tasks using the DCE had to position the endo-scope with both hands, lock it, and then release one



hand to manipulate the first instrument to pick up the ob-ject. Once the object was grasped, the subject had to ma-nipulate the endoscope again to allow the secondinstrument to touch the object to transfer between hands.In this process, a single operator struggled to perform ef-fective bimanual coordination while holding the platformand maintaining a steady surgical view. In fact, the DCEmodel had the second longest performance times re-corded in this study (140 � 55 seconds). Adding operatorsto manipulate the endoscope or work instruments hasbeen used in medical practice as a temporary solutionto this problem. For example, when performing humantransgastric cholecystectomies, the primary surgeon hasto control the operating platform, adjust the endoscopeconstantly to view the surgical site, and give instructionsto assistants to manipulate the surgical instruments.Such a solution requires an increased mental and physicalworkload, which stresses the primary surgeon under thiscondition and may lead to increased error.

Second, the instruments used with a standard endo-scope or R-Scope lack the degrees of freedom to make in-dependent movements around the surgical site. Currentendoscopic designs allow instruments to move forward

Figure 4. Task performance of 3 operating systems: R-Scope, 160 � 71

seconds; DCE, 140 � 55 seconds; DDES, 29 � 28 seconds (P ! .001).

Figure 5. Task performance of expert and novice group using 3 operat-

ing systems.


and backward, rotate, and open/close with 3 degrees offreedom. Adding elevators, as the R-Scope does, gives 1additional degree of freedom for each working channel.This would seem to be a feasible solution for instrumentinteraction. However, our data did not show a perfor-mance advantage when using this design (160 � 71 sec-onds). Both experienced endoscopic participants as wellas the novices actually lost time trying to use the elevators.This was probably because of the complexity of the con-trols and the inability of the instruments to truly interact.

Third, the linked movement between the endoscopeand instruments causes serious problems for bimanualcoordination. Bimanual manipulation requires indepen-dent movement of the instruments and endoscope. TheDDES is designed specifically to allow separate movementof individual instruments as well as the endoscope. The ben-efit of this design for complex procedures was confirmed inthis study (DDES: 29 � 28 seconds vs DCE: 140 � 55seconds and R-Scope: 160 � 71 seconds; P ! .001). Thisresult supports our hypothesis that separation betweenvision and motion, increased degrees of freedom of theinstruments, and an ergonomic user interface are beneficialfor performing complex bimanual surgical tasks.

Another benefit of a stable and ergonomic rail platformis that only 1 operator is required to perform an entireNOTES procedure, much as a surgeon would performa laparoscopic procedure. The primary operator can fullyconcentrate on the surgical manipulation once the surgi-cal site is in view. A DCE always needs at least 2 operatorsto perform basic surgical tasks, which is inefficient, createsergonomic and communication issues, and has potentiallyhigher surgical costs.

This study also has implications for designing the nextgeneration of operating systems for image-guided andminimally invasive surgery following the guidelines pub-lished by Natural Orifice Surgery Consortium for Assess-ment and Research. The laparoscopic paradigm, asembodied by the DDES, has advantages compared withthe endoscopic paradigm; however, the laparoscopic par-adigm may not be the ultimate manifestation of future de-signs. An intuitive paradigm built around an endoscopicplatform with robotic technology could be even more fa-vorable. A system that combines the advantages of flexibleendoscopy (small diameter, retroflexion ability, and theflexible shaft) with the manipulation capabilities of con-ventional or robotic laparoscopy could open the surgicalfield for widespread adoption of NOTES. Of course, me-chanical systems such as the DDES are appealing froma cost standpoint and certainly represent a good solutionuntil the true place of NOTES in the realm of patient careis determined.

Limitations of our studyWe compared bench-top testing only. We had to use the

operating devices in the order in which they were avail-able in our laboratory. The better performance of the

www.giejournal.org


DCE than that of the R-Scope could be explained by theorder of the tasks. The R-Scope tended to be tested first,the DCE second, and the DDES third. For example, tricksthat had been learned with the R-Scope (such as bendingof the graspers to transfer the ring) may have been laterused with the DCE. This order did not affect the DDES be-cause the graspers work differently from their endoscopiccounterparts.

CONCLUSION

Currently available endoscopes present some inherentdifficulties when used for complex surgical proceduressuch as NOTES. New designs seek to solve the problemsby increasing the independent motion of the instrumentswith regard to the image. We found, however, that theadded complexity of devices such as the R-Scope often off-set the advantages of independent instrument motion.The best results were found when a design offers indepen-dent instrument movement, separation of vision and in-struments, and a stable ergonomic interface.

REFERENCES

1. ASGE/SAGES Working Group on Natural Orifice Translumenal Endo-

scopic Surgery White Paper October 2005. Gastrointest Endosc 2006;

63:199-203.

2. Swanstrom LL. IRB Requirements for Human NOTES Procedures

SAGES. Philadelphia, PA, 2008.

3. Zorron R, Filgueiras M, Maggioni LC, et al. NOTES. Transvaginal chole-

cystectomy: report of the first case. Surg Innov 2007;14:279-83.

www.giejournal.org

4. Marescaux J, Dallemagne B, Perretta S, et al. Surgery without scars: re-

port of transluminal cholecystectomy in a human being. Arch Surg

2007;142:823-7.

5. Bessler M, Stevens PD, Milone L, et al. Transvaginal laparoscopic cho-

lecystectomy: laparoscopically assisted. Surg Endosc 2008;6:6.

6. Pearl JP, Ponsky JL. Natural orifice translumenal endoscopic surgery:

a critical review. J Gastrointest Surg 2007;5:5.

7. Zorron R. Update on South American experience with NOTES chole-

cystectomy in humans. Paper presented at: Digestive Disease Week

2008; May 17-22, 2008; San Diego, CA.

8. Bardaro SJ, Swanstrom L. Development of advanced endoscopes for

Natural Orifice Transluminal Endoscopic Surgery (NOTES). Minim Inva-

sive Ther Allied Technol 2006;15:378-83.

9. Swanstrom LL, Whiteford M, Khajanchee Y. Developing essential tools

to enable transgastric surgery. Surg Endosc 2008;22:600-4.

10. Kalloo AN, Singh VK, Jagannath SB, et al. Flexible transgastric perito-

neoscopy: a novel approach to diagnostic and therapeutic interven-

tions in the peritoneal cavity. Gastrointest Endosc 2004;60:114-7.

11. Pasricha PJ, Hawari R, Ahmed I, et al. Submucosal endoscopic esoph-

ageal myotomy: a novel experimental approach for the treatment of

achalasia. Endoscopy 2007;39:761-4.

12. Kantsevoy SV, Hu B, Jagannath SB, et al. Transgastric endoscopic sple-

nectomy: is it possible? Surg Endosc 2006;20:522-5.

13. Swain P. Nephrectomy and natural orifice translumenal endoscopy

(NOTES): transvaginal, transgastric, transrectal, and transvesical

approaches. J Endourol 2008;22:811-8.

Received June 3, 2008. Accepted December 23, 2008.

Current affiliations: Minimally Invasive Surgery Program, Legacy Health

System, Portland, Oregon, USA.

Reprint requests: Georg O. Spaun, MD, Minimally Invasive Surgery

Program, Legacy Health System, 1040 NW 22nd Ave, Suite 560, Portland,

OR 97210.

If you want to chat with an author of this article, you may contact him at

[email protected].


mailto:[email protected]

bimanual coordination in natural orifice transluminal endoscopic surgery: comparing the conventional...

Documents