Endolumenal Surgical Robots vs Endolumenal robots : The challenges

As promised I’m continuing to get deeper into what I think will be part of the future of surgery over the next decade. Let me recap my own personal history in this space to give context.

In 2007 I laid down a series of patents in an emerging filed that was know as NOTES (Natural Orifice Translumenal Endoscopic Surgery) and NOS (Natural Orifice Surgery.)

Both are similar in the fact you avoid cutting the skin - and you go into the body either transoral or transanal. In NOS you stay inside the lumen of the body and operate internally. In NOTES you puncture the lumen and head off out into the abdomen or chest and operate like “laparoscopy.”

Back in the day, myself, Brad Sharp and Wayne Noda created a company called MINOS medical (Minimally Invasive Natural Orifice Surgery  - proud of that one.) And we set upon a few novel treatments by wanting to bring similarities to cardiology to the bowel. Endolumenal interventions using long catheters, overtubes and long flexible endoscopes. Rapid exchange was a big part of the concept.

We actually did quite a bit - including an appendix inversion and removal from inside the colon. Plus a treatment system for diverticulum from inside the left side colon.

We were not alone - there were several companies using endoscopes and advance instruments to do things like ESD (endoscopic submucosal dissection) of sessile polyps. Endolumenal suturing (Endostich by Apollo endo surgery.) And there were a host of companies (big strategics and start ups) looking into this - including Olympus, JNJ and more.

Why am I boring the pants off you with this?

Well in 2006 / 2007 we were all making a similar approach - and all hitting the same brick wall. You see robotics was still in its infancy and really only in lap surgery - and mainly the prostate. There were some novel systems starting out that were a bit “hybrid” like Medrobotics (rest its soul) and Transenterix with systems like Spider.  But there was not a real flexible “robot” trying to do what we were attempting within the lumen of the bowel.

To be fair Transwenterix did also develop Surgibot.

Hitting the endolumenal wall

No I’m, not talking the physical bowel wall (but we’ll come back to that) - we are talking the virtual wall all companies hit by using manual flexible endoscope technology. You see they are created for diagnostics, and limited to “snaring and grabbing.” So they have functional limitations with imaging - instruments approach and “drive”. They are flexible - and not designed for doing “surgery” as such.

So let me start by dissecting an endoscope. You basically have a long flexible tube that has a series of “channels” inside it and a set of cables to help steer it.

It has to be flexible enough to be able to bend - but rigid enough to be able to be pushed and pulled to advance and retract without breaking it.

It has a basic imaging system that is fixed at the tip of the scope - and then some working channels for air, water, and some limited flexible instruments such as snares, needles, or baskets.

The issue is that you are seeing from the end of the scope in a linear axis, so have a very “parallel” view to the lumen of the bowel. Great to see ahead but not brilliant to look up or down.

The next issue is that any instrument pops out in a linear - parallel fashion to the scope and the imaging. And that movement is very constrained to mainly forwards and backwards (I’m paraphrasing.) but it is not steerable in the sense of a robotic wristed instrument.

So if you want to move it up or down (of course you can rotate it through torsion) you move the tip of the scope up and down - and everything is fixed relative to it… so the image moves up or down.

And you only have one “arm” instrument in most cases. So with that one linear instrument (that really is not that strong) you are limited to what you can do (try tying your shoe laces with a weak floppy stick with just one hand.)

So Endoscopes are very good at diagnostics and limited therapeutics (taking off mushroom like polyps with a snare) or grabbing stones - or injecting etc etc.

In the hands of the “best” that can do way more, and even some ESDs and other things. But it is very hard and only for the brave.

If you try too much and you create a hole in the bowel (you go too deep) or get a bleeder - then your choices are limited to be able to repair. There are things like OVESCO (over the scope clips) and other devices - but it’s more emergency recovery than purposeful intervention.

So as we, and the pioneers, tried to do advanced surgery inside the bowel or even translumenal in the abdomen (let’s not talk sterility just yet) we all hit this technological wall. The limitations of vision and instruments and control of the scopes just started to stack up.

What is funny now is that I remember sitting in our board room with our engineers and saying “What we need is a fucking robot.” Because thats how I roll… I state the obvious. Well done me.

What did we need?

If we wanted to do serious procedures inside the bowel - or the stomach - we needed to return to the basic paradigms of surgery.

We needed a few things to be able to do something “more like laparoscopy.”

Firstly we needed an off axis camera - you can’t do serious interventions inside the colon if your camera is in line with the delivery mechanism - the endoscope.

So we really needed a camera that could be advanced and then articulated up so you could get a panorama view of the bowel - and the WALL. Now we talk about the bowel wall - because that is where a lot of the action will happen.  The camera must be independent of the delivery system -  with independent movement from the “delivery tube” (or scope in flexible endoscopy.) I can’t explain how critical that feature is.

Next - most surgery requires traction and counter traction - and for that you need two arms. Two instruments so you can do push and pull. And what was clear was that the had to NOT be parallel to the delivery scope. The instruments  had to come out and then elbow outwards so they could reach back centrally - or give triangulation - as it’s known. Then they had to be independent with independent movement so you can pull one - push another. Spread tissue apart - hold things up while working on the other end of the tissue.

(Robotics for Natural Orifice Transluminal Endoscopic Surgery: A Review - Scientific Figure on ResearchGate. Available from: https://www.researchgate.net/figure/The-Cobra-triangulating-scope-USGI-Medical_fig13_258384895 [accessed 5 Dec 2024])

Now to be fair to our team - we did work this out with a system called Megachannel. This was a massive overtube that you used with a standard colonoscope to deliver in the bowel - and then like rapid exchange in cardiology - we pulled out the scope and inserted a flexible channel strip. A tube with multiple channels. We would then use a small 5mm scope through the channel that could pop out and be moved independent of the megachannel. We had two nitinol (memory shape) tubes that came out and created elbows that allowed triangulation. It was cool because it allowed you to use off the shelf instruments like small graspers.

But all of this was relatively passive and weak and without real control. In our experiments we understood you needed to be able to drive the arms and scope , drive them independently and have a degree of fine control. And it had to have very strong gripping forces.

We also understood something fundamental. If both hands were used to just control the scope - then you needed other users controlling the instruments - plus advance and retract etc etc. It was clear that a single operator would not be able to do this. And that impacts reimbursement and profit. (A story for another day.)

Finally we understood that clipping the bowel wall closed was not going to work. It was clear we needed suturing (especially for closing the musocsa in an ESD) - and it needed to be fine suturing with needle holder capability. And that required a lot of strength on the closure of the jaws, and being able to drive the sutures through tissue.

Of course you see where this is going… we needed a robot. And well sadly for us… we had other fish to fry in cardiology and hernia so we stopped there.

Endolumenal robots

In the time since I was doing manual endolumenal approaches a lot has happened in the world of robotics. There have been many companies that have thought about all the things we identified as needs - and then how to pack them up into a robot… why?

Well put simply - if you get an Endolumenal robot - and attach it to a console - you end up with removing much of the complexity away from the pilot. You can advance the endoscopic robot - move the camera independently - have small articulated arms that you can control from the console.

A console driven endolumenal surgical robot creates a control paradigm that greatly drops cognitive load - and sets up for a more “surgical” laparoscopic (forget that word in it’s true sense) paradigm.

I’m making major differences here between endolumenal robots and endolumenal surgical robots. The surgical capable robots are a sub category of endolumenal robots. And we need to start to break down from simple “diagnostic” - “Diagnostic that deliver energy therapies” - “See n treat robots” - to “Fully fledged surgical endolumenal” robots.

It’s a graded spectrum that steps up complexity and functionality as you move across that map.

You start with just making it easier to navigate the flexible robot scope - so the core function of either ION by Intuitive, or Monarch by Johnson & Johnson, or Noah medical.

Their core functionality is about using smart navigation - route planning - and robotic manipulation of pitch - yaw and roll - to get a tip that you can control with precision - in a simpler way as you navigate say - the bronchial tree.

This allows the scope to get deeper into the lung - faster - and more directly (follow the GPS). When there - they can then do a needle biopsy of a lesion. So in effect it’s diagnostic endolumenal robot.

That would be the same moral equivalent as say Pietro Vladastri’s magnetic diagnostic colonoscope. It uses a form of robotic steering to get a scope to go and see things and maybe do simple biopsies. (See my other post.)

Let’s go up a level - and think about now post biopsy, bringing a laser fibre, or microwave technology down the robot’s working channel in the lung and delivering energy - so a “Diagnostic” that starts to allow a targeted energy therapy to ablate the tumours.

You still do not need multiple complex articulating instruments to do this - you need a needle through the lumen of the bronchoscope - drive the bronchoscope to the lesion and prod the energy by advancing the device. It’s basic - and is relying on the movement of the whole robotic tip - not independent articulation of the instrument.

Let’s go up another level. We now deliver a flexible robot and start to advance through the working channel some more advanced instruments like graspers, baskets, clips - etc. And this single arm (because you have small diameter limited scopes for say the urinary tract) has some independent manoeuvrability to the endoscope. It may have some flex - pitch - yaw - grasp. So you can start to do basic treatments beyond just energy application. Think of stone removal - or interventions up in the Kidney, or small lesions in the bladder.

And that may come as PCNL - (percutaneous Nephrolitotomy) in the case of Monarch.

It will be “basic surgery” and require more complexity of drive at the operator end (Monarch went from 2 arms in Bronchoscopy to 3 arms in renal applications so they could drive the suction irrigation Percutaneous catheter with the third arm.) - I’ll do a whole post on this as an adjunct to my rendezvous post.

But you can see this is a gear change in complexity. However the camera is still the scope - the arms are limited in size - strength and functionality. The laser fibre is still linear in how it is delivered by the scope.

Let’s take the next rung up the ladder. Now if we want to do some “surgery” inside the bowel - we are going to need a serious robot. As said we need independent vision and camera control to get the camera distance from the working area without moving the whole delivery platform. Then you need to deploy a minimum of two working arms - fully articulated that can do things like lift tissue - triangulate - retract - suture etc etc. And all independent of the delivery platform (which is now no longer an actual endoscope.)

Of course we have already seen this configuration in the single port systems from Intuitive - and Titan. They deliver shorter - rigid instruments that just have flexible tips. The system is “delivered” by a rigid delivery system through an abdominal port. So although in some ways “similar” to flexible robots in conceptual delivery of a flexible camera and flexible arms  - it differs enough in complexity once you try to do all that over a much longer distance and with completely flexible instruments for their entire length.

So it is major step up even from an SP… and a step change in complexity from a flexible endoscopic diagnostic robot.

We are also seeing forms of hybrid technology come in the shape of systems like Virtual Incision and their MIRA. Mini robots that are part rigid and part flexible scope and arms.

Flexible Endoscopic Surgical Robotics Platform

I’m now going to focus on the one platform that has been pioneering this field of larger format - fully capable endoscopic surgical robots - Endoquest.

They have very kindly furnished me with some information, as I think their journey and progress is a great lesson for us all in startups of robots. They have given me the warts and all inside of their struggle to get where they are today. And as I sit and write this - they have just announced their approval of their IDE - a big milestone for flexible endoscopic robots. (Not the first in man- they’ve done that in the past OUS.)

I want to first talk the history of Endoquest - to highlight a few of the challenges that bringing such a complex platform has. Inherently trying to get a robot to work completely flexibly from the operator end to the business end (in the patient) automatically makes the whole platform unstable, hard to control and just generally weaker than a rigid laparoscopic robot. Even an SP.

If you know about robotics you know how hard the control algorithms are for rigid instruments with wrists - and rigid arms with well defined joints. Now think about trying to do that for a fully flexible delivery system - with independent flexible instruments - and a flexible camera to boot.

So as stated above - getting a flexible delivery system - and then at least two flexible instruments and flexible camera to all work at distance is in itself a control system nightmare. But there is one thing on top of that we have not dived into. To do diagnostic - or supply a therapy catheter - it is very hard as it requires to actually”clean” and  “sterilise” the system. The complexity of the electromechanical system is a nightmare from a cleaning and sterility point of view.

(Residual dirt endoscopes - Cori L. Ofstead, Abigail G. Smart, Krystina M. Hopkins, Harry P. Wetzler, The utility of lighted magnification and borescopes for visual inspection of flexible endoscopes, American Journal of Infection Control, Volume 51, Issue 1, 2023, Pages 2-10, ISSN 0196-6553,)

In fact - most colonoscopes today achieve only high level disinfection - because they are not doing fully surgical tasks - and they are in an already “dirty” non sterile tract. But if we want to break the blood barrier in a significant way - or even go full thickness and enter the abdominal cavity - then that high level disinfection has to become sterility. (How long it stays sterile is a question once in the colon… but…)

And it is not just “Sterilising” the system (you can do that with Sterrad etc) - it is actually the cleaning and decontamination that is the hardest part. These are highly complex tubes with cables, wires, channels and a host of small moving parts where dirt can get stuck. Every cm of length increases the complexity and difficulty in cleaning, decontaminating and then sterilising. Big issues for any design team.

So when you put all this together - (I know I looked at it in 2008) - the mountain to make a fully functioning - sterile - endolumenal - surgical robot is quite staggering. And I think that is why we are nearing 20 years since NOS and NOTES before we see the first human cases with a full flexible endolumenal robot. It is way harder than a soft tissue lap robot.

Let me discuss Endoquest

The company was founded in 2016 - but not as Endoquest - it was founded as ColubrisMX. This was the first generation technology spun out of the Microsurgical Robotics Research lab at the University of Texas medical school in Houston. Behind that technology and convinced to make it work was Dr Todd Wilson. In the background was an investor that stuck with Endoquest - Mr Hamid Jafar. It takes people with vision to push this sort of technology and not give up. The company entered a first IDE as Colubris…and was set to do robotic ESD. Now I don’t have all the answers here but the data being generated (didn’t seem to be safety or efficacy issues) looked like it would not support the complex DeNovo pathway - so that study was abandoned. And this is one of the things that robotic start ups will go through - all of them. From the actual first clinical use of the system there is a huge amount of learning and insight that comes to light. It can be around things like usability, cleaning, sterilisation, durability.

I want to scream - there is a massive gulf between a clinically ready robot - and a commercially viable robot. We all need to get that into our heads- they are not the same thing.

Stepping into my startup role in life here. Hey this shit is hard. But having an MVP in clinical cases is where you learn the mistakes - make the pivot and come back stronger. It must be safe - but being in humans is a critical step to perfecting towards a commercially viable product and commercially viable support system / model.

After this setback, Colubris became Endoquest and a generation 2 flexible endolumenal surgical robot was developed - but based on all the knowhow from the  Colubris MX learnings.

Fred Moll stepped in on the funding round and added his weight to this company - and a new management team were given the console to drive this thing. (Bit of a pun.)

Not for today, but I’m going to dig deeper into Endoquest as part of my series of “Making robots is hard. Lessons learned.” So watch out for that one.

Until companies like INEN (out of IRCAD) reveal their system - or JNJ do a Monarch Endolumenal surgery robot  (go see my Auris posts) - or Intuitive launch something - I only have this one example. But I will bet this is the archetype design for all colonic robotic systems.  And I think they have a pretty hefty IP portfolio around this. ( I think about 150 patents on last count.). But more so - coming back to those complex control algorithms - I am sure that “secret sauce” is going to give them clear water for quite a while.

So let me describe their system and see if it ticks all the boxes I made from back in 2008.

The whole thing by some sources is said to be called the ELS system. I think that means Endolumenal Surgical. At the centre of it is a flexible delivery system - that they call EndoDrive. In effect it is a long flexible - steerable over-tube. An electromechanical driven set of channels within a… well  tube. Not much bigger than a classic endoscope 18mm . (Colonoscopes are about 13mm max).

And that EndoDrive attaches to the bedside cart - where it can be advanced and retracted once in place. Now to see where it is going it has a camera within the EndoDrive system. Of note is there are channels for insufflation and off the shelf tools - just like any colonoscope.

Remember to make a working space in the colon and not cause a vacuum in the bowel - you need to pump air in to “insufflate the bowel” - just like colonoscopy.

So let me be precise here - most of the driving the system to the actual target site would be quite long and slow if it was done on the robot. Plus the drive mechanism would have to have a very long travel distance - making the cart unruly.  So Endoquest took a very practical approach - where the EndroDrive is manually driven (like a colposcope) to the target area - ( I think they may change this for the future as Autonomy comes along… but…)

Once at the target area the EndoDrive is then attached to the bedside or patient cart.

Down one channel is the endoscope - which now gets attached to the bedside cart.  And down two other large channels (6mm) two long flexible - steerable instruments are placed.

From this point onwards the console operator can control: the snake like camera- the two instruments - and can drive the EndoDrive in and out and move it around (flex it) to always be in the optimal position of the working site.

The instruments are twin jointed 7DOF instruments - don’t expect them to drive like a daVinci wristed instrument - think of them more as an SP style of instrument. They are resposable and multi life - which has been no mean feat. The company have had to invent their own proprietary cleaning system to get the debris off prior to sterilising. Instruments are often the hardest part.

The camera is an interesting “snake like” 2D camera  - and why that is important is two fold. First you get a top down view to see an operative field better. Second it is independent of the drive system so even if you have to flex the drive you can maintain your surgical filed of view. Okay I lied - third - and final for today. You need snake like cameras to move the light source up and out of the way not just the camera lens. That is became glare and shadows are critical in any surgical field. And keeping the light just at the level of the instruments does not optimise the field illumination.

I said 2D - and everyone may be saying - but 3D is best. Yes - but you have to look at the limits of the physics. This camera must be small. The minimum diameter of any working 3D is about 5mm and it’s not great. Better is 8mm and best is 12mm. Because it allows you to get a distance between the two camera lenses (inter pupillary distance) to get that spacial operation to give good 3D. In these small systems - it’s difficult to get good 3D and a great 2D is sufficient for these depths of fields etc. Maybe as they go lap - that will change. But hey Virtual Incision seems to be doing fine with a 2D camera.

So now this answers many of the issue I found back in the day at MINOS Medical. And it answers those key barriers. It brings better visualisation - brings triangulation - 6mm instruments bring strength and retraction; counter traction and suturing capabilities. In effect it brings the things you need to do small endolumenal surgery.

I know I’ve talked a lot about Colonoscopy - as that is there they are going first but an 18mm OD drive system opens up - upper GI, thoracic, transumblical, transvaginal - so who knows. It could potentially compete with some of the single incision systems like SP.

But I think today it fits neatly into the endolumenal… where as of today it carves out its own space. At the moment it is out in there lead. But I know that all the other companies will be charging along behind. But never underestimate first mover advantage.

A note on Intuitive SP

I know a lot of people at Intuitive will be reading this and grinding their teeth saying “But SP already is used in transanal and transoral surgery.” So yes technically they are doing endolumenal surgery of a sorts. It is back of the mouth - or rectal surgery. I think we can all agree that I’m talking here about much deeper into the tract. Getting around flexures and needing a fully flexible endoscopic robot. But is does scream at me SP and ION will have a love child; and it will be Intuitive’s endolumenal robot. If they are not already in cadaveric testing I’ll eat my hat.

But until we see something beyond patents - I merely speculate.

What does this potentially do to competition?

The first thing about endolumenal surgical robotics is that there is no big company owning this space today - Like Intuitive does with lap robots. It’s blue ocean. But you have to think about who this could massively impact…

It could impact the endoscope providers - so Olympus, Pentax, FujiFilm. If these smarter endolumenal robots get installed - then they could start to become more and more the “weapon of choice” in the more complex cases. They will certainly start to be the thin end of the wedge in procedure numbers erosion (where the profit is) - and that means that people need less scopes over the years and less service agreements.

I do also think that this then becomes the trojan horse for the Intuitives, the JNJs, the Medtronics to get into this lucrative endoscopy space. They come in with high end disruptive product - and they build imaging capabilities, they build infrastructure for service - they build sales teams and commercial capabilities. They start to create “bundles”. Well it’s a small step to then use those systems for straight diagnostics. Add on smaller scopes - add on more diagnostic drive systems and controls. Bring in AI and automation… and well… they are in that space and the classic Japanese scope companies will have three new monster competitors.

Don’t believe me - just look at bronchoscopy today. We now have JNJ and Intuitive as the talk of the town in bronchoscopy. And Monarch is already making waves in urology - with high end PCNL procedures. It’s coming folks.

And YES - look at the Auris original thesis - JNJ are all over this space once they get Ottava sorted. It’s a branch of Monarch meets Ottava - and it completes their offering.

Same for Medtronic - they didn’t just buy Fortimedix for shits n giggles.

And ION by Intuitive - mixed with the SP know how is a slam dunk for this space.

First mover advantage is going to be huge here !!!!

If I was in the scope company business  — I’d be either buying up companies like Endoquest. Or investing in emerging endolumenal companies now  - or getting hardcore into the build to buy model and get a project going at full speed. Full full speed.

For the likes of Boston Scientific - they produce a lot of widgets that a competent suturing system will negate. Could this finally be the area where Boston Sci decides they need to venture into this robotic space?

For all the robotics companies - this future war will be built around capability bundles of “systems” that can work in multimodal ways - meet up in rendezvous - share data - make dual approaches - give optionality - and be in the treatment pathway at all stages from diagnostics to treatment. None of the big three are going to want to hand off a patient to another competitor to have them do the treatment. You do that by also owning the diagnostic. See n Treat - and therapy.

One key capability in the next 5 to 10 years will be the ability to do endolumenal - endo organ surgery. Companies like Virtuoso Surgical will also start to become interesting tuck ins. It is going to be a war of who can aggregate the most technology around them - and then give a single unified look / feel and even drive system. It will be about getting hospitals into their 360’ wrap around ecosystems.

Just as the surgical robotics space is segmenting - I believe that the endolumenal robotics space will also now see a segmentation - and part of this segmentation will either take surgical procedures from the OR surgical robots - augment those procedures of surgical robots with rendezvous - or delay the final surgical procedure through less aggressive earlier treatments. For sure they are going to bring new levels of accuracy, precision, ease of use, and data.

It’s a very exciting time and area.

These are just the opinions of the author for educational purposes only. All images, trademarks, brands are owned by their respective companies.