3D-Visualization Improves the Dry-Lab Coronary Anastomoses Using the Zeus™ Robotic System

(#1999-9522 ... October 25, 1999)

Helmut Gulbins, MD, Dieter H. Boehm, MD, PhD, Hermann Reichenspurner, MD, PhD, Martin Arnold, Reinhard Ellgass, Bruno Reichart, MD

Department of Cardiac Surgery, University Hospital Grosshadern, Ludwig-Maximilians University, D-81366 Munich, Germany

ABSTRACT

Introduction: Robotic surgical instruments enable quick and precise movements and may allow complete endoscopic coronary artery bypass grafting. However, cardiac surgeons will have to become familiar with this technology and endoscopic viewing. We present our training program with special focus on 2D- and 3D-visualization.

Methods: A thoracic skeleton, covered with a neoprene suit, served as model for the chest wall. Either a glove, fixed on a metal plate, or a pig heart were placed inside for training. On the glove, a suture line consisting of two lines of 16 points each, with a distance of 2 mm between each point, was stamped. On the pig heart, the LAD was prepared and incised; subsequently an anastomosis was done using the dissected right coronary artery as a graft. The time required was measured for both models. For suturing, the Zeus™ System (Computer Motion, Goleta, CA) was used and the third robotic arm positioned the endoscopic camera. The scopes were connected to a 3D-camera and the picture was displayed on a headset with two integrated monitors. Visualization was set to either 2D or 3D. Three surgeons were involved in the study. Each one did at least 12 anastomoses on 2D and 3D.

Results: The three surgeons involved showed a clear and rapid learning curve. The times required for the suture line decreased from 12.5 ± 1.6 to 8.5 ± 0.5 minutes with 2D and from 11.9 ± 5.4 to 7.8 ± 0.5 minutes for 3D respectively. This decrease did reach statistical significance (p = 0.03). In the pig heart model, the anastomosis times decreased from 33.2 ± 8.4 to 15.7 ± 0.3 minutes with 3D-visualization, and from 36.2 ± 2.2 to 29.5 ± 3.3 minutes with 2D. The decrease in anastomosis time did again reach significance (p = 0.025). At the end of the study, the times achieved with 2D-visualization were significantly longer than those with 3D (p = 0.01).

Conclusions: A surgical training program is mandatory to become familiar with these new technologies. Both models showed learning curves over an acceptable time course. 3D-visualization facilitated quick and precise movements, thus resulting in shorter anastomosis times.

INTRODUCTION

Minimal or less invasive operation techniques are an expanding field within cardiac surgery. For coronary artery bypass grafting, different techniques have been developed, depending on the patient's individual state and demands [Calafiore 1996a, Calafiore 1996b, Nataf 1996, Borst 1997, Calafiore 1997, Mack 1997, Nataf 1997, Diegeler 1998, Reichenspurner 1998, Shennib 1998]. However, complete manual endoscopic bypass surgery remains a goal that has not been achieved yet. Robotic surgical instruments and telemanipulators may allow the surgeons to overcome the technical problems [Garcia-Ruiz 1997, Stephenson 1998a, Shennib 1998, Stephenson 1998b, Shennib 1999], as they enable quick, precise, and scaled movements without tremor. A second handicap within endoscopic surgery is the 2D-visualization, as it takes additional concentration towards the orientation and away from the main operation field. We present our training experience with a robotic telemanipulator system (Zeus™; Computer Motion, Goleta, CA), with special regards to the improvements using 3D-visualization.

MATERIALS AND METHODS

A chest phantom, suited with a nontransparent cover of neoprene, was placed on an operation desk. Three ports were placed at the third, fourth and fifth intercostal space for the robotic instruments and the scope. The instrumentation ports were introduced at the anterior axillary line, the camera port at the mid-sublavian line. With this port placement, the instruments and the optic formed an equilateral triangle. The scope was connected to a voice-controlled robotic arm (Aesop®, Computer Motion Inc.). For endoscopic suturing, the Zeus™ telemanipulator system (Computer Motion, Goleta, CA) was used [see Figure 1 :349:].

Movements of the surgical instruments are controlled via handles that resemble conventional surgical instruments [see Figure 2 :351:]. The instrument movements are scaled and any tremor is filtered, allowing accurate microsurgery. The instrument positioner controllers form the link between the surgeon console and the instruments. These controllers analyse the input data measured at the surgeon console as the surgeon moves the handles. The instrument positioners are then directed to move appropriately. This ensures that the instrument tips duplicate the movements of the surgeon handles at the console. The input data can be scaled by the computer, as the analogous link between the surgeon and the instruments has been replaced by a digital one.

For visualization, a small 3D-camera (Vista®, Westborough, MA) was used. The distance from the camera to the pig heart or suture row was 4 cm at minimum and 5 cm at maximum. The camera was connected to a video-tower (Vista®) for image processing. The 3D-images were faded to two small LCD-monitors integrated in the head set (Vista®). Visualization could be switched from 2D to 3D easily; therefore the surgeons always wore the headset and did not use the additional 2D-image on the normal screen, even if working in 2D.

The three surgeons in the training program consisted of an experienced cardiac surgeon, a cardiac surgeon who had completed the training program, and a "raw trainee". All three had had a short training program of five anastomoses with the system to become familiar with the telemanipulator. These five anastomoses were done using different sutures.

Two objects were used for training. A simulated suture row was drawn on a glove using a stamp. The distance between the single points was 2 mm., with 16 pairs forming the complete suture row. The glove was fixed over a metal plate and put inside the chest phantom. After installing the telemanipulator system, a double-armed, 7 cm long suture was placed on the glove and suturing was initiated. It started with 4 knots for initial fixation of the suture; thereafter the anastomosis was done in a running fashion through all points. At the end again 4 knots were done and the time required was measured.

The second model was used to simulate coronary bypass grafting. Hearts from pigs were obtained from the local slaughter house. After the LAD was prepared and incised, the right coronary artery was harvested and used as a graft. Anastomosis was done with a running suture, starting at the proximal edge of the incision (see Movie). Each surgeon completed at least twelve anastomoses with the system. At the end, 6 knots were set and again the time required was measured. The quality of the anastomosis was checked by injection of blue ink via the graft.

Different materials were used for suturing. All sutures were 7 cm long and double armed. The materials were either 6/0 Gore-tex® (W.L. Gore & Associates, Munich), 7/0 Fumalen® (Fumedica, Herne, Germany), 7/0 Prolene® (Ethicon, Hamburg, Germany).

For statistical analyses, the times required for an anastomosis with 2D- or 3D-visualization on the pig heart were compared using paired student's t-test. For the suture line model, 2D- and 3D-visualization were also compared using paired student's t-test. Statistical significance was assumed with a p < 0.05 on a 95% confidence interval. All values are declared as mean value ± standard deviation. For analysis of the learning curve, the required times for the first three anastomoses (or suture lines) were compared with the most recent three anastomoses (or suture lines) using student's t-test.

RESULTS

Suturing with the 6/0 Gore-tex® suture was easier than with 7/0 Prolene®, as it showed nearly no memory effect. The Gore-tex® sutures also were more resistant to mechanical stress, thus facilitating their use. This resulted in slightly shorter anastomosis times (29.4 ± 0.8 minutes versus 31.9 ± 0.7 minutes, both at 2D), although the differences did not reach statistical significance (p > 0.05). On the other hand, 6/0 Gore-tex® sutures were more traumatic to the tissue, due to the size of this suture. Additionally, the needle was not that sharp, so the forces for penetrating the tissue were higher, increasing the suturing trauma to the vessels. Fumalen®, as a compromise, proved to have less memory and 7/0 sutures, with a sharp needle. We therefore chose the Fumalen® suture for the further training program comparing 2D- and 3D- visualization.

All surgeons involved showed a clear and rapid learning curve during their training program on the two models. With 2D-visualization, the times required for a suture line decreased from 12.5 ± 1.6 to 8.5 ± 0.5 minutes [see Figure 3 :352:] (p = 0.03). The corresponding times for 3D were 11.9 ± 5.4 to 7.8 ± 0.5 minutes respectively [see Figure 4 :354:] (p = 0.02). The suturing times for the first three anastomosis with 3D-visualization did not differ significantly from those achieved with 2D. The average times required for a running suture on the glove model were longer with 2D-visualization (8.7 ± 0.57 minutes) compared to 3D-visualization (8.0 ± 0.5 minutes). However, these differences did not reach statistical significance (p > 0.05).

At the pig heart model, there was no statistical difference between 2D- and 3D- visualization regarding the required times for the first three anastomoses. The suturing time decreased with 3D-visualization from 33.2 ± 8.4 to 15.7 ± 0.3 minutes [see Figure 5 :355:] (p < 0.05). The corresponding times with 2D were 36.6 ± 2.2 to 29.5 ± 3.3 minutes [see Figure 6 :356:] (p < 0.05). The difference between the initial and final time required for an anastomosis was more pronounced with 3D-visualization (17.5 minutes) than with 2D (7.1 minutes). The average anastomosis time was 30.7 ± 0.9 minutes with 2D-visualization and 16.9 ± 0.8 minutes for 3D. This difference did reach statistical significance (p = 0.01).

The times achieved with 2D and 3D were compared in figure 7 for one surgeon. The graph clearly points out the similar times for the first anastomoses and the more pronounced learning curve with 3D-visualization, resulting in shorter anastomosis times at the end of the study. When testing the anastomoses, in two of the first five training cases of each surgeon, a technical error was detected (n=6). In five cases a major leakage occurred, making additional stitches necessary. In one case, no flow through the graft was seen due to a closure of the LAD. From the sixth case on, no more problem at the anastomosis site occurred.

The subjective impression of the surgeons involved was that 3D-visualization improved the co-ordination between the right and left instrument. This facilitated the handling of the needle. Especially the transfer of the needle from one instrument to the other was easier with 3D, reducing the forces applied to the needle and suture. With 3D-visualization the anastomoses were done in a way comparable to conventional procedures, although moving speed remained slower than on usual cardiac surgery (see Movie). All three surgeons reported 2D-visualization to be more strenuous on the eyes than 3D.

DISCUSSION

Robotic instruments are thought to be essential for endoscopic bypass surgery, as they reduce the tremor and enable the precise movements required for microsurgery by scaled down motions [Garcia-Ruiz 1997, Shennib 1998, Shennib 1999, Stephenson 1998a, Stephenson 1998b]. However, cardiac surgeons are not all familiar with endoscopic instruments, thus requiring a training program for these new techniques. With our models in the dry lab, the surgeon had no direct view on his instruments, as the phantom was covered by nontransparent neoprene. Therefore, he had to rely completely on the pictures received through the endoscopic camera. The first step, suturing on a glove, helped develop the general feel for the instruments. Doing an anastomosis on an isolated pig heart was the next step, requiring more sensibility for the system. With this model, suturing on a complex 3D-surface was practiced. It demanded the ability to compensate for the loss of force feedback by the optic perception of tissue alterations, tension of the sutures, and the movements of the instruments.

The loss of feeling at the finger tips and, therefore, the lack of force feedback can only be compensated by optical impressions. However, surgeons accustomed to the technique felt as if they were sensing the forces in their fingers. Of course, this was only a subjective impression as the brain transforms the optical perceptions into tactical impressions. This translation will only be effective when there is a constantly sharp picture on the screen. Every loss of dissolution of the images will surely lead to reduced perception of small movements and structures, thus reducing these indirect tactical impressions of forces. Interestingly, all three surgeons showed quite similar learning curves during their training program despite their different experiences with coronary artery surgery. At the end, all three did the anastomoses in the same times within a small range. This indicates that these techniques, using a telemanipulator, are something new even to experienced surgeons.

Coronary vessels and the corresponding bypass grafts are commonly within the range of 1 - 2.5 mm. Therefore, the surgeon requires a 4- to 6-fold magnification to recognize all important details and to be able to do exact stitches. The color, as well as the black and white, contrasts are therefore very important to give an exact image. The 2D-view of normal scopes requires additional brain work, as the orientation in space is only possible using indirect markers, such as shadows, size of instruments, or movements of different structures. This is more fatiguing to the surgeon than the normal physiologic 3D-view he is used to. With new technologies, 3D-visualization has become possible in endoscopic surgery also. The two pictures, sent through the two small monitors integrated in the head set, form a 3D-picture to the viewer. Our data proved these 3D-pictures to facilitate endoscopic suturing and manipulation. As the orientation in space was much easier with a 3D-impression, the movements of the instruments became faster and more precise. The cooperation between the right- and left-handed instrument improved with such procedures as exchanging the needle from one instrument to the other. These improvements resulted in faster anastomosis times when compared to 2D.

Stephenson et al. [Stephenson 1998b] reported average anastomosis times of 31.7 ± 2.0 minutes in a similar pig heart model with 2D-visualization, but Shennib et al. [Shennib 1999] also reported faster anastomotic times when using 3D-visualization and endoscopic instruments. These data agree with our mean anastomosis times of 30.7 ± 0.9 minutes with the 2D-model. However, the times required for an anastomosis with 3D-visualization were shorter in our study, underlining the advantages of 3D-visualization technique. We did not see significant differences between 2D and 3D with the glove-model of a suture line. This may indicate that the advantages of 3D-visualization are more pronounced in real complex 3D-objects than on a plane surface. Therefore, an additional 3D-model with an irregular surface for the training program should be developed.

The suture materials play an important role in endoscopic bypass grafting. Regularly used sutures were too long, as the instruments would leave the image each time they tightened the suture. Short sutures of 7 cm length were used because such long distance movements are uncomfortable and motions outside the camera window risk hurting the patient. The so-called "memory effect" of the Prolene® and Fumalen® sutures was the main disadvantage of these materials, as knot tying was more difficult compared to the Gore-tex® sutures. However, Gore-tex® sutures were available at 6/0 only. Additionally, the needles of Prolene® and Fumalene® were sharper and therefore more suitable for anastomoses. The lack of force feedback demands a sharp needle, thus reducing the energy required for each stitch, thereby minimizing the related tissue trauma. However, Gore-tex® sutures will soon be available in 7/0 and, with a new needle, will possibly combine the advantages of the sutures used in this study.

The learning curves for all three surgeons involved in our training program indicate clearly that 3D-visualization facilitated endoscopic manipulations. The times required for an anastomosis at the end of the training program were still longer than those in conventional cardiac surgery. An important desirable element would be a sensible force feedback for direct control of the power applied through the instruments, with further improvements of visualization, especially more color contrast and depth of focus. All these issues may reduce anastomotic times further and make suturing more comfortable.

CONCLUSION

Using the Zeus™ telemanipulator system (Computer Motion, Goleta, CA), endoscopic bypass anastomoses are possible. Suturing utilizing 3D-visualization facilitates the anastomoses, thus resulting in a faster learning curve and shorter anastomosis times. A training program is mandatory to practise endoscopic suturing and to become familiar with the new technology of telemanipulation in combination with 3D-visualization.

REVIEW AND COMMENTARY

1. Editorial Board Member KK138 writes:

a) This paper describes the results of some surgical tasks that are performed under various vision conditions, namely 2D vs 3D. It is well known that performance increases under 3D vision as compared to 2D vision provided that the same spatial resolution is provided for both viewing conditions. The finding of the authors is therefore not new. There is extensive literature on this matter, none of which is mentioned by the authors. Since this is a vision study it would be of interest to the reader what the exact spatial resolution of the vision system was.

b) Considering the setup, it is important to know what the actual distance to the object was. It is known that incongruent axes (i.e., misalignment between operator viewing direction and controler orientation) impair operator performance. Since a headset mounted display was used, how was this problem approached?

c) The biggest problems with vision studies is that besides resolution and depth perception, motor skill is usually tested as well. It is however important to seperate these different skills because they independently influence test results. It has been pointed out that performance time is only a poor measure since it also is affected by dexterity to a great extent. Especially when testing a new device the learning curve doing the task is paralleled by the new experience of working with a computer enhanced instrumentation system.

d) In addition, for small group sizes the t-test does not apply at all.

e) The authors state that for analysis of the learning curve,the required times for the first three anastomoses (or suture lines) were compared with the most recent three anastomoses (or suture lines). One look on the figures shows that the obviuos outliners (Fig. 3 datapoint 3, surgeon 3) do not allow for this comparison. In Fig. 4 there is only a difference for surgeon 3, while surgeon 1 and 2 perform equally at the beginning and the end of their series. The conclusions drawn from this comparison is therfore not justified.

f) The dots in Figures 3-7 should not be connected by lines as this is not appropriate. Figure 2 adds no new information and could be ommitted.

g) Somewhat unclear is the impact of the various suture material that was used during the study. It is confusing to read that three types of needles were used at various stages of the experiment, but it is unclear when those changes were made.

h) Statements on "forces" should not be made at all since forces were not measured. The discussion again does not focus on the existing literature of the influence of vision systems on task performance.

i) Speculative statements on needle forces should not be made (no measurements were performed).

Authors' Response by Helmut Gulbins, MD:

a) Our intention was not to report something completely new about visualization but to report on our experience with different visualization options using the Zeus™ system for endoscopic coronary bypass grafting.

b) The distance between optic and subject was 4 cm at minimum and 6 cm atmaximum. We have added this to the text in "Material and Methods".

c) In our series, three different surgeons at different skills (as mentioned above) were involved. All had no experience with either 2D- or 3D-visualization in endoscopic surgery. All had an introduction to the system doing five anastomoses with 2D prior to the series to get used to the telemanipulator. We have added this latter sentence to the text.

d) We used the paired student's t-test, testing paired date (first anastomosis with 2D vs. 3D, second anastomosis 2D vs. 3D, third ... and so on) due to the small group size. We think this is adequate.

e) With the rather small groups, it is correct that this comparison isstatistically not completely correct. But we compared 2D with 3D and showed,that both visualization modes required a learning curve. The differencesbetween the first and last anastomoses were compared; however, we did notshow a difference between the learning curves with 2D and 3D. The commentwould be correct, if we drew from these comparisons the conclusion that themeaning curve with 3D was faster than that with 2D. But we did not; atleast only the end points of the anastomoses times differed between bothmodes and these differences were significant. If you wish, we can remove the comparison between the first and last threeanastomoses from the text.

f) We know that from a mathematical view it is not correct to connect the datapoints. However, we think that the presented graphs clearly point out thelearning curves we intended to show. Graphs with just data points usually arenot that clear. We therefore want to maintain the design unless the reviewer insists on changing it.

g) The different sutures were tested in the first anastomoses that we did to get use to the system. We tried to find out which suture type was most convenient. We have added this information to the text.

h) We think that one can talk about forces without measuring their absolute value. However, if a suture is ruptured, the forces applied were too high.Higher resolution and different vision systems do influence task performance, but at least there is no real tactile feed-back and, in our opinion, this is the main issue.

i) The statements would only be speculative if we mentioned values for the applied forces. We stated only that forces can be applied at a level that causes damage either to the suture material or to the tissue. This is not speculative as we could see the results.

AUTHOR/ARTICLE INFORMATION

Reprint requests to: University Hospital Grosshadern, Ludwig-Maximilians-University Munich, Marchioninistr. 15, D-81377 Munich, Germany, Phone: 089-7095-6465, Fax: 089-7095-8873, Email: H.Gulbins@hch.med.uni-muenchen.de

Submitted on: October 22, 1999; Accepted on: October 25, 1999

REFERENCES

1. Borst C, Santamore WP, Smedira NG, Bredee JJ. Minimally invasive coronary artery bypass grafting: on the beating heart and via linited access. Ann Thorac Surg 63:S1-5, 1997. :9203587:

2. Calafiore AM, Giammarco GD, Teodori G, Bosco G, D'Annunzio E, Barsotti A, et al. Left anterior descending coronary artery grafting via left anterior small thoracotomy without cardiopulmonary bypass. Ann Thorac Surg 61:1658-63, 1996a. :8651765:

3. Calafiore AM, Angelini GD, Bergsland J, Salerno TA. Minimally invasive coronary artery bypass grafting. Ann Thorac Surg 62:1545-8, 1996b. :8893612:

4. Calafiore AM, Teodori G, Di Gaimmarco G, Vitolla G, Iaco' A, Iovino T, et al. Minimally invasive coronary artery bypass grafting on a beating heart. Ann Thorac Surg 63:S72-75, 1997. :9203603:

5. Diegeler A, Falk V, Krahling K, Matin M, Walther T, Autschbach R, et al. Less-invasive coronary artery bypass grafting: different techniques and approaches. Eur J Cardiothorac Surg 14:S13-19, 1998. :9814786:

6. Garcia-Ruiz A, Smedirna NG, Loop FD, Hahn JF, Miller JH, Steiner CP, et al. Robotic surgical instruments for dexterity enhancement in thoracoscopic coronary artery bypass graft. J Laparoendosc Adv Surg Tech A 7:277-83, 1997. :9453871:

7. Mack M, Acuff T, Yong P, Jett GK, Carter D. Minimally invasive thoracoscopically assisted coronary artery bypass surgery. Eur J Cardiothorac Surg 12:20-4, 1997. :9262076:

8. Nataf P, Lima L, Regan M, Benarim S, Pavie A, Cabrol C, et al. Minimally invasive coronary surgery with thoracoscopic internal mammary artery dissection: surgical technique. J Card Surg 11:288-92, 1996. :8902643:

9. Nataf P, Lima L, Regan M, Benarim S, Ramadan R, Pavie A, et al. Thoracoscopic internal mammary artery harvesting: technical considerations. Ann Thorac Surg 63:S104-106, 1997. :9203611:

10. Reichenspurner H, Guliemos V, Daniel WG, Schüler S. Minimally invasive coronary artery surgery. New Engl J Med 336:67-68, 1997. :8984335:

11. Reichenspurner H, Boehm DH, Welz A, Schmitz C, Wildhirt S, Schulze C, et al. Minimally invasive coronary artery bypass grafting: port-access versus off-pump techniques. Ann Thorac Surg 66:1036-40, 1998. :9768998:

12. Shennib H, Bastawisy A, Mack MJ, Moll FH.: Computer-assisted telemanipulation: an enabling technology for endoscopic coronary artery bypass. Ann Thorac Surg 66:1060-3, 1998. :9769003:

13. Shennib H, Bastawisy A, Mcloughlin J, Moll F. Roboter computer-assisted telemanipulation enhances coronary artery bypass. J Thorac Cardiovasc Surg 117:310-3, 1999. :9918973:

14. Stephenson ER, Sankholkar S, Ducko CT, Damiano RJ. Robotically assisted microsurgery for endoscopic coronary artery bypass grafting. Ann Thorac Surg 66:1064-7, 1998a. :9769004:

15. Stephenson ER, Sankholkar S, Ducko CT, Damiano RJ. Successful endoscopic coronary artery bypass grafting: an acute large animal trial. J Thorac Cardiovasc Surg 116:1071-73, 1998b. :9832700: