Restoration Robotics - Robots for Hair Transplantation

 Another tedious and cumbersome job is being taken over by a robot - hair transplantation. Restoration Robotics, a venture backed company in Mountain View, CA,  has developed a robot that can harvest hair follicle units (FUE) from a person's head in a minimally invasive way. The company received FDA clearance and is planning to start selling the robot this summer. The robot solves many of the difficulties that the restoration procedure has had and it may well expand the market for hair transplantation.

 Here is a transcript of the conversation with Mohan Bodduluri, Vice President R&D and one of the founders of Restoration Robotics. It's important to note that the company has an all-star board of directors  and investors - the Who's Who in the field of medical robotics - and that they are also working on a future product that will take care of the hair implant process as well. (This interview was originally conducted for a robotics column on the Wall Street Journal Japan.)

 

Q. How did the company come about? 

A. The idea of using a robot in hair transplantation was originally that of Dr. Phil Gildenberg who is a neurosurgeon from Houston. One day he was having dinner with his young dermatologist nephew who discussed with him how hair transplantation in those days - and even today - was a tedious and difficult procedure. So Dr. Gildenberg says to himself, hey, I’ve stuck a lot of probes into people’s brains using robotics and image guidance, why can’t I put a few follicles in someone’s head? So he wrote a patent application, submitted it to the patent office and then brought the idea to Accuray which was started by a fellow neurosurgeon.

  I happened to be at Accuray at the time as Vice President of R&D, and so he and I met. We thought it was a great idea but Accuray was focused on making its first product CyberKnife successful and starting a brand new product in a totally different field did not make sense. Eventually Accuray became more successful and bigger and then I decided to move on and to start this company. (Photo: Dr. Bodduluri with a prototype robot.)

 Accuray did not invest in our company but because the idea got kicked around there first, we wanted to make sure that there were no issues later. So we made a clear agreement with Accuray and as part of the agreement they became a minor stakeholder in the company.Q. Who are the other 3 co-founders besides yourself? 

A. There’s Dr. Gildenberg who invented the idea. And Don Caddes who was my boss at Accuray at the time, who also served as CEO at Accuray for a while. Before Don and I started looking for money, we were introduced to Dr. Fred Moll who is a serial entrepreneur and medical robotics legend. He is the founder of Intuitive Surgical and other successful medical device companies. He liked the idea and once he said that he will help us raise money, it became very clear that this was going to happen. At that point, Don and I left Accuray and founded the company.

Q. What is the technology behind all this? 

A. Obviously we need a robot. It has 6 degrees of freedom. We needed some kind of visualization so we need some cameras to see the hair so that we can guide the robot to the right location. And very early on we decided that we need stereo vision because you need 2 eyes to able to see the depth and how far the distances are. We ended up using 4 cameras - 2 sets of 2 eyes. One set of 2 eyes looks at a very small area, maybe few millimeters. The other set looks at a bigger area maybe about 5 cm square. So the cameras that look at the smaller area are very precise and they can visualize the hair very accurately and they help the robot get to the hair very accurately. The other camera is able to help the software and algorithms to look ahead and see where the robot needs to go – on other words to plan ahead. For example, I don’t want to take out every hair, only every third hair.

(You can watch a video of how the robot works here.)

  To remove the follicles, we need needles. And once we have the needles we need to actuate them and to move them in and out and turn. So we designed a mechanism to do that. The disposable needles were developed here. The cameras can see the needles as well as the hair so that they can get the needle to the hair very accurately. The cameras are wired all the way back into the robot so when the cameras see the needle and the hair are slightly off it tells the robot to move over and so on. And it’s doing it in a continuous basis. In the robotics field this is known as visual servoing. That is considered a very advanced field and I’ve only seen it in universities and people doing research. I would say that this is probably the first product that uses this technology.

Q. So basically, the head might move and you want to be sure you are getting to the right position. 

A. Exactly. So what happens is if you make calculations and you say hair is over here and you ask the robot to go from here to there, even if the head did not move, there’s usually an error. By the time you get there it might be off, but if there is a way you can put feedback there and say how much farther, then you can incrementally get closer. So that’s what we are doing using the cameras. All the tracking is being done with the camera that is looking at the smaller area.   

 And then we have to take all this equipment and put it into a box. So we made a nice looking cart so all the equipment is inside here so we just wheel this inside the office, plug it in and it’s pretty much ready to go.

 Also it turns out when you put a patient on a standard table, they breathe and when they breathe their head moves. When you are looking at a small area of hair, a hair is 100 micron wide, and the patient moves a millimeter or two, it looks like a gigantic move. And to make it more comfortable to the patient to provide some kind of stability, we came up with this custom chair that was made so the patient can sit there in a comfortable fashion. It isolates the chest from the head a little bit more than when you lie down. And with a special chair, you can have the patient in a similar location every time so that the robot doesn’t have to work too hard.

  Cameras can see hair but that’s not enough. The computer has to understand what is hair, what is a follicle, etc. You have to be able to distinguish these things automatically. And not only in one image but two images and you have two totally different pictures of hairs, a couple of hundred hairs here in the left image, a couple of hundred hairs here in the right image, and you need to know which one is which. And you need to figure out the depth. All that needs to be done automatically. And then the patient is moving from one video frame to the next video frame and you need to know where all the hair went. We can track about a couple of hundred hairs in 3 to 5 milliseconds on a continuous basis.

Q. Once the patient comes in, what is the process?

A. The patient comes in and they first trim his hair down to about a millimeter and then anesthetize locally. And then there is something called the “tensioner” that you put on the patient’s scalp. The skin moves a lot so the tensioner is to create a little bit of a tension by stretching the skin so that it’s easy to punch through. Once we do that, you drag the robot over and press a couple of buttons and then it’s pretty much ready to go. And once you say go, it automatically selects a hair and the needles are actuated automatically and keeps doing it till it does the whole field.

Q. How large is the area that the robot is taking care of?

A. The tensioner is about 5cmx5cm. You don’t want to clear the whole thing because then you’ll have a bald spot. There are parameters that you can adjust, but the default does about 15-20% of the hair. 

Q. How fast does the robot do the work and how does that compare to a human? 

A. If you let the robot go non-stop, it can do 1,000 follicles an hour. But since you have other things to do, including moving the tensioner, in reality it can do about 500 an hour.

As for human beings, there is a lot of variability. There are some doctors that claim that they can do 500-700 an hour by hand, or even 2,000 an hour. Maybe they do it for 5 minutes and multiply that number.

 The important thing here is that, most doctors don’t do this, and that the doctors that do it, hate to do more than 500 a day. So if you need about 2,000 follicles, most doctors who offer this will do it in 2 or 3 days. Whatever they take out that day, they put it in that day. Most doctors don’t even offer this procedure, which is called FUE (follicular unit extraction).

 The common way to do it is called the strip technique. The comparison really is with this. Basically you go from ear to ear. It’s about 30 cm long, 1cm wide, and 1cm deep. They remove that strip from the scalp and then the technicians sit in front of their microscopes and slowly cuts them carefully all the way down to these follicular units with razor blades. And each follicular unit is typically one hair to 4 hairs. The cutting takes 15 minutes, stitching takes 15 minutes, so the doctor is done in 30 minutes. The patient is sitting around while these technicians, half a dozen of them, sit there and cut for another 3-4 hours. So it takes about 4 hours to cut 2,000 grafts. If you don’t want that big cut, then you want to go with this FUE technique, and if you go with FUE, it take 2-3 days to do it, so here we are saying have the robot do the 2,000 in the same 3-4 hours but with no cut.

 

(Photo: Controlling the robot is simple, just a few clicks to start.)

Q. What is the cost? 

A. Famous doctors charge a lot more, but the average cost for the strip technique is about 5 dollars a graft (=follicular unit). If you do 2,000 it’s about 10,000 dollars for the procedure. For the FUE, since the doctor has to do everything, they will charge approximately 10 dollars per follicular unit. Now these numbers are different in Japan. I believe it’s about 8 dollars per graft for the strip and FUE is about 12 dollars per unit. In Japan a very few number of doctors do this.

Q. How much does your robot cost?

A. The system is 200,000 dollars in the U.S. And then on top of it, there’s a “per use charge” of 1 dollar per graft. We will provide disposable needles free of cost. We have a secure system, called ARTAS KEY, that enables the physicians to purchase “clicks”, and the system keeps track of the usage.

Q. When will the physician recover the cost? 

A. It depends on their volume but it’s actually very quick. A doctor who does not know even how to do this procedure (FUE), you’re making 5 dollars per FU and as soon as you get the robot, you can start charging 10 dollars each and then you give us one dollar. You are certainly making 4 dollars extra on the same patient. So your money comes back very quickly, It can be as soon as 6 months, maybe 9 months. You still need to hire technicians for implanting, but you won’t need as many. 

Q. What is the reason that it can only be used for black and brown straight hair?

 A. When you start off, you want to start with the easiest case, and it’s easier for the camera to see if there was a contrast between dark hair on lighter skin. Those are the patients that we did the clinical study on. Curly hair like mine may turn a lot more inside the scalp than straight hair, and that may cause a little bit more damage to the hair when harvested, so at the time of the clinical study we had not studied all that. But since then we have figured blond hair, red hair, we have done curly hair including my own. It can be done. It’s just that we have not done enough studies, that’s what the indication says on FDA, but it’s up to the doctors to decide. We can’t claim in our literature that it can be used on other patients until we have enough studies showing that, and we are doing the studies right now, so we may be able to say that pretty soon.  

Q. Can you talk about how you got FDA clearance?

A. Yes. We had a long conversation with the FDA for the past many years. Finally we reached something called “an understanding” of what needs to be clinically shown for the clearance to happen. And at that time we were supposed to do a statistical study – to statistically prove that the hair harvested by the robot will survive the same way, same percentage as manually harvested hair follicles. That was the goal. And we came up with a protocol that said, we had to do 36 patients in 2 different clinics. Each patient was transplanted 60 hairs – not follicular units – by the robot and by the human. We took picture and then we waited 9 months after the transplant and take pictures again in the area we transplanted. The FDA requirement was that we had to show that we were at least as good as the humans, in statistically terms it’s called a “non-inferiority study”. And that’s the study we did.

We started in June 2009, and by the time we treated all the patients it was Dec. 2009 and then we had to wait for 9 months. We had a third party count the number of hairs. And we did the statistical analysis and submitted in Nov. of 2010. We received the clearance in April 2011. After we submitted it, it was very quick, but the length of time it took for us to do the study is unusually large for such a simple procedure like this. But that’s because maybe the FDA is more concerned about esthetic devices. 

Q. So was the result better than humans? 

A. Statistically we can’t say that. The way you do the calculations, you can’t say that you are better because statistically not only there is an average, there’s also the noise. If the noise is this big, it’s hard to say that this one is better than this. I can’t even say the word equal. I have to say “non-inferior”, because statistically all these things mean differently. But what I can say is that the robot seems to do better in not damaging the follicles. We’ve improved the technology even more since the FDA study. 

Q. When do you plan to start shipping the robot? 

A. We are planning to ship out the first system in July. And the first year, we want to focus on the U.S. market, probably the West Coast. Perhaps sell about 10 systems because we want to be very deliberate making sure that we provide all the support and service for doctors so that they can learn, treat a lot of patients and publish. We want to ramp up slowly. 

Q. How large is the market? 

A. Very difficult to say. It turns out that just in the U.S., it’s believed that there are about 100,000 to 150,000 hair transplant procedures per year.  90 plus percent are strip, and because the average cost is 10,000 dollars, it’s about a billion dollar market right there. But if you look beyond that, there are about 30 to 40 million bald men in the U.S. So the market size is really huge. Hopefully with the technology, we can interest many more men to actually come in for hair transplantation. In the past, LASIK came along and a lot of people got it because of the technology. We believe that the same thing can happen and the market can easily go to 2x, 3x, even 10x. Even then it’s a very small compared to the total number of bald people. 

Q. How are you viewing the Japanese market? 

A. The Japanese market is a big market, but very few procedures are being done. But we know that there’s a company called Aderans which is very successful, in wig business. So there is a demand on the consumer side for a solution. We also know that the Japanese like technology and this minimally invasiveness. 

Comments

[Rena Graham]

This is really one of the big leap when hair transplant surgery is concern. I wonder though, which of which is more costly?

[roboty]

So how many % of men have black or brown straight hair? That chair looks very uncomfortable.