“We can take your workload online for you and you’ll now have access to over 200 pathologists.”
“You’re going to get dedicated pathologists that your clinicians will get to know personally, because through our LIMS they can talk to them directly about cases and share information.”
Conversation with Alistair Robson
Consultant Dermatopathologist and Clinical Director of Digital Pathology at Source LDPath, London. U.K.
BIOSKETCH: Dr Alistair Robson is a qualified general (FRCPath) and specialist dermatopathologist (DipRCPath), and was a Consultant Dermatopathologist at St John’s Institute of Dermatology, St Thomas’ Hospital, London, for 14 years. He is engaged in clinical research and has published over 100 articles, mostly concerning various forms of skin cancer. Dr Robson. has over 15 years’ experience of teaching dermatopathology to both dermatologists and pathologists. He was the founder of LDPath and is now Clinical Director of Digital Pathology at Source LDPath, a company that provides rapid histopathology reporting by specialist pathologists to over 330 hospitals in the U.K. and abroad and is on a journey of helping pathology departments digitize and incorporate AI into their processes.
Interview by Jonathon Tunstall – 3rd Feb 2023
Published – 24 May 2023
JT – Hello Alistair and welcome to Pathology News.
I see from your bio that you are a career dermatopathologist and clearly very well-published in this field. You’ve also spent your career researching various forms of skin cancer and I think it’s fair to say you are a world-renowned expert in this field. In 2011 you founded LDPath, now Source LDPath, an organization, which we’ll hear more about later on.
Perhaps I could start by asking you something about your early career and what brought you specifically into the domain of digital pathology.
AR – Well, in my early career, of course, digital pathology didn’t figure, so my early career was very conventional, just training in general histopathology and the realization that within that field, dermatopathology was an area I particularly enjoyed for a variety of reasons to do with the discipline.
I also liked clinical academia and interestingly, I had a lectureship for a while at Oxford. One of my colleagues, also a lecturer, was researching in the very early stages of what was called telepathology, which was then about viewing digital images, or a representation of microscopic images, through the medium of a screen. That was in its early days and, all of us, just because we were in the same room, ended up taking part in a couple of in-house trials. I just remember thinking this was very ingenious, but I didn’t pay much attention, to be honest. So, my career moved forward in the conventional way, and I secured a consultant position at St. John’s, which is a well-known and renowned dermatology hospital in London, with its own in-house pathology department. I enjoyed that experience, and at that time, I started building something of a private practice, as all consultants do. Ultimately this led to me founding my own company, LDPath.
Anyone who works in the NHS as a doctor knows the frustrations of the NHS very well. They perhaps differ from specialty to specialty, but they’re legion. I could see this, and it was frustrating. You couldn’t effect change, you couldn’t pioneer. There were so many vested interests, power groups, and barriers. The NHS had started to move away from the old-style consultants who ran their own departments to consultants who gradually became employees, and it was very much the administrators who ran things. Of course, there were a variety of reasons for that which brought along both advantages and disadvantages, but it all added to a feeling of emasculation, of being unable to effect change within your department. There were also other downsides to the NHS. The slowness and the lack of response to what was required were frustrating. Conversely, working in the private sector, it was fast, and it was responsive. I didn’t own a company back then, but I worked for private enterprises. I could see that they could make changes much more quickly and if they wanted to alter an aspect of the service, they could just do it. There was nevertheless a cultural perception that the private sector didn’t have the probity of the NHS, so there were separate advantages and disadvantages (or at least perceptions of those things) to both the public and private sectors.
The point I’m trying to make is that this annoyed me, and I thought, “why does it need to be?” Therefore, in founding LDPath, which was non-digital in the beginning, I wanted to offer a very high standard of service with the same probity as the NHS, with specialist reporting, specialist cases, but also being responsive to the client’s needs with a fast turnaround time and high quality. I also wondered (although this has not yet played a major role at Source LDPath) why the state sector should have a monopoly on teaching. We actually did do some educational things in the early days, but it’s not something that we have yet really developed. The point was to build a system that could provide the best of both worlds, and I started that in the background of my private practice. The vision was that this could serve as a model for how pathology is done in the UK.
JT – It sounds as though you were very much catering to what you saw as a need in the market at the time.
AR – Yes, I think that’s right. I mean, this is always the case with any historical perspective. It’s very easy to look back and telescope a narrative but it’s fair to say that a variety of things became apparent to me over this period and fell into place. It’s a mistake to say everything was planned exactly as it has turned out, but I thought, “well, why can’t we do this? There is a need for it, for something a bit different and innovative and for building a reporting system that is purely for pathology.” That is because in most cases in NHS hospitals, they will have a reporting platform which is a one-size-fits-all for microbiology, for clinical chemistry, for haematology etc. When a new software system is introduced into an NHS hospital, representatives from various departments come along and they give their opinion. Inevitably you end up with a system that is okay for all, rather than any single department thinking it is fantastic. The difference with our system is that we are bespoke just for histopathology.
JT – Well, in 2011, I remember LIMS systems were quite ubiquitous at the time, but it was still quite early days for whole slide imaging in tissue pathology, and certainly for the quality and reliability of digital scanners.
AR – Well, we weren’t scanning back then, but I do remember a particular watershed moment. I was giving a talk in either Switzerland or Germany. I just remember the room and who was talking. The guy who was speaking said, “I’m going to present my cases digitally through an Aperio scanner.” For me this was a watershed moment because I was at the back of the room and I looked and thought, “I can diagnose these from here. Wow!” This must have been around 2000, but in the early days we had no commercial impetus to go digital. There were other things we were concentrating on. It was not until four or five years later that I seriously considered going digital and that was driven to some extent by personal circumstances as I was moving abroad.
This did however help me to make the decision to go abroad, and then we basically pioneered using digital generally. Over the next few years, the real crisis in the UK pathology service began and digital proved to be not just convenient, but we could have pathologists reporting from home which had huge commercial potential and went some way to solve the problem of an insufficient number of pathologists. So, for me, the two came together very nicely, and the rest, as they say, is history.
JT – It’s interesting listening to you speak there because I was with Aperio around 2010, and 2011, and at the time, the main uptake was in the pharmaceutical industries for drug discovery. Companies like Novartis, AstraZeneca, GSK etc all had a lot of whole slide scanners at different world locations, all connected together. But it was the very early days of digital in clinical pathology. You described that eye-opening moment when you stood at the back of the room and thought, aha, this technology is here, and it does sound like that was quite a watershed moment for you.
AR – It was, which is why I remember it so well. I can’t actually remember whether I was in Germany or Switzerland at the time, but I can remember the person, it was Reinhard Dummer from Switzerland and I pretty much only remember that. I don’t even remember what I was speaking about myself or much about the conference. I just remember that moment and thinking, my goodness, this is really good.
JT – So let’s talk a little more about LDPath, or more correctly Source LDPath because you’re now part of Source BioScience, correct?
AR – That’s correct. We joined them last year for mutually beneficial reasons which are certainly being realized, but the growth we enjoyed prior to that was also considerable. We did a number of things and made a number of commercial decisions which I think were correct and are still being pursued. One is, and this is a key point, our LIMS system is bespoke, so our reporting system is written by us. There’s no off-the-shelf system to deal with and you are quite right, there were a number of off-the-shelf systems available in the days before I started LDPath. I hunted around. I had a couple of potential business partners and we looked at some off-the-shelf platforms from America. They’ve got some great systems, but I remember meeting one particular representative and he said, “well, the implementation costs will be anything between fifty to a hundred thousand and up to 2 million, depending on the extent you wish to be connected.”
Well, I was a sole trader, at that time looking to become a commercial entity and one hundred thousand is a lot of money just for a reporting system. In any case, I was already familiar with some reporting systems, and I found it frustrating that they didn’t do what I wanted them to do and there was no way to alter that. So, we went down the route of coding our own. I’ve made some good and some bad decisions over the years but that was a really good decision because now we own it, we can adapt it, and we are bespoke to histopathology.
Now, this is what I always impress upon the team. They are there to code things to make my life easy. They have to be absolutely responsive to what I need as a pathologist. The analogy I give to them is when I went to work in a hospital, I’d open the door and go into my office, and I’d have my microscope there, I’d have my phone there to talk to the technician, and I’d have my tray for cases I need to discuss with my colleague down the corridor and perhaps, he or she wants to discuss an opinion with me etc. This is the office, that a pathologist knows, with a tray for stuff coming back from the laboratory with some extra cuts or IHC or whatever. I want to enter the LIMS and see the same setup as I do in the office of a pathologist.
JT – Well, this is the difference in building your own system, isn’t it? Your new features are not being coded by IT geeks who think these are cool features. You’re doing it because these are useful and standard features of the histopathology workflow.
AR – Correct. We’ve got great coders and they’re very nice people, but they see things in a certain way. The user, of course, sees things from his own perspective and we always get frustrated when a new adaptation comes out, because we can’t even speak their language. Recently for example, they put together a ‘Precisions Document’ and you think, what the hell is that? They know what it means, but we don’t, so what we’ve now done, is to get it coded so that we can actually change it at the business end. Because of this, we now don’t have to worry about them. They do what they want and if they get the front wording wrong, we just have to clarify a little more what they mean. The analogy would be that you don’t get into your car and have a switch saying, ‘take fuel and put it in the compressor,’ it just says ‘ignition.’
So, we’ve developed our LIMS so that the front end is written by a pathologist. Our coders code it brilliantly, and it works very well but then we add some guidance to adapt it for a pathologist seeing this for the first time and to give him or her the feeling that they are in a familiar environment. Now we can say “here are your cases, this is where you can put your holidays, this is where you send to that lab.” This type of system is unique in the UK.
JT – So tell me something about Source LDPath as an organization. You’re obviously a dermatologist, but I guess you handle quite a range of different disciplines, different disease areas. I’m wondering about the service as well, you provide scanning services and you also provide image analysis, so how does it all bolt together?
AR – Well the principal workflow is a diagnostic service. We’ve got two workstreams actually. We’ve got one where wet tissue comes in, gets processed in a laboratory, put on a scanner, and then a pathologist will read it, do whatever they need to do in terms of additional work, and ultimately produce a report. Then we also have distressed NHS trusts sending us their backlogs of pre-prepared slides which we scan and send off to our pathologists. They are of course, this being an online service, all over the country. If you take any average histopathology department, there are some particular big volume specialities; skin is one, gynaecology is another, GI is another and then breast and urology are also fairly big. Those disciplines account for about 90% of your average histopathology department. We cover all of them and because we have around 200 pathologists on our books, we cover any other specialty. We pride ourselves on that, on covering any specialty. That is with a small caveat, as renal, neuropathology and ophthalmology are so specialized that, although we handle them, they are not part of our bulk system right now. We should be able to handle those in due course.
JT – That’s good, and as you say this I’m thinking about your wider role in the market. It seems that you are in fact providing services for those labs that don’t want to go down the digitization road themselves for whatever reason. Is that a correct perspective? The build it or buy it scenario, digitize or outsource?
AR – I think that’s right, but there are other factors at play here. One is the extreme distress that many NHS trusts are in. They can sometimes have a bottleneck of several months of unreported cases. They are also having to deal with their KPIs and the metrics they are being held to in terms of cancer wait times etc. That means the trusts are often quite desperate to find a solution.
Simultaneously, there is a growing awareness of in-house digitization as a potential solution to these issues. However, I would say that some departments, don’t always connect the dots. For example, if you’re a department of three pathologists, and you should have six pathologists, you’re in distress because you’ve got a mounting backlog. The department then decides to go digital and they get a scanner in, but it doesn’t help them, because they’ve still got three pathologists, not six. There will be some efficiencies because perhaps the pathologists can now stay at home, don’t have to commute two hours a day and can report more cases more easily. Ultimately though, you can’t load balance, because you’re still one department with three pathologists. What we say to those departments is “if we come in, you won’t have three pathologists, but now 203, and if you appoint somebody new, you can simply reduce the amount that we report for you. Equally, if someone leaves, retires, goes sick, takes maternity leave, you can give more work to us. Your, pathology recruitment headaches are over because you’ve taken it online so that you can now load balance across the country.”
Understanding that solves one problem, but there are also other things you have to bring in to create a premier service. Conventional outsourcing for example. In a conventional histopathology department, most of the reporting is routine, but a significant number of cases per week are not and the pathologist needs to have a discussion with the clinician. This is a really important aspect of patient care, but it’s not only about patient care because the very nature of the relationship you have with a clinician becomes important. You come to understand each other and work together in interpreting clinical pathology because pathology is an interpretation, it’s not just a “result”, which some people forget.
Now when you outsource in traditional outsourcing models, you as a clinician might receive a report from a doctor you don’t know who is 300 miles away. Your hospital staff are happy because a KPI has been satisfied, but you are thinking that the report doesn’t make sense and you don’t really know the pathologist. It is hard to emphasize how critical that relationship is. So, our solution is that we will take your workload online for you, but we’re not giving it to just anyone. You’re going to get dedicated pathologists that your clinicians will get to know. They will get to know them personally because through our LIMS they can talk to them directly about cases and share information.
This is what I mean when I talk about taking the role of a pathologist and their office online. It’s not just about KPIs. It’s not about turning around a report just to satisfy a cancer wait time. It’s about the delivery of a premier service. When we do take it online, we are solving the NHS’s problems around the country, but we’re always conscious of maintaining the key relationships. We’re also continually developing and improving. We have our new LIMS 2 coming out soon and I’ve got a whole list of things that I know we need to do, areas where we can still make improvements. That I think is the big thing that marks us out from the competition.
JT – That’s great. I love your approach and your LIMS platform is clearly a major component of the new digital ecosystem that we see emerging in the field of tissue pathology. I can’t help wondering about the future of labs based purely on traditional microscopy, because you can imagine a time in the not-so-distant future when most young pathologists are trained on digital systems. Will they be willing or even able to do their work using a microscope and could it become impossible to recruit into a non-digital environment? What’s your view on that?
AR – It’s an interesting point. My own view is that probably there won’t be a role for the microscope at some point. Likely we’ll end up with some residual little niches, for example, coronial work, and autopsies, but that’s not diagnostic work in the conventional sense. I do actually think at some point somebody will come along and address these areas, but this is further downstream.
In terms of training, that’s one of the things we’ve got to implement. We’ve got the systems in place on the current LIMS to do it, but we haven’t adapted it just yet. To explain in a little more detail, all these departments need to meet their KPIs, but they’ve also got juniors to train. When I was a registrar, the consultant would give me several trays of work and say, “go and look at that. Write your reports, then come back, and show me.” Then we would go through on a doubleheader looking at the slide together and, they would say, “yep, agree, or nope, you’ve missed this, or you should think about this.” This training technique is still in use today, so, we are coding our LIMS so that the consultant who gets the allotted cases digitally can send them to the junior and they just work together digitally.
There are several advantages here. First, of all, if a junior writes their report and it’s a nasty malignancy, they don’t have to wait two days to return the whole batch that they’ve finished. That one can immediately come back to the consultant. Also, we’ve designed it so they can see and annotate the slide together, so the consultant pathologist can agree or point out something missed, and the junior can ask questions. One other thing that can be built in, is the consultant being able to state when the slides should be returned, and if they’re not done by that allotted time, they’ll then revert automatically back to the consultant.
We can also build image repositories. Over the years, I’ve built up several of those which must now total 40 or 50 thousand images. I can use any interesting cases to teach. I might for instance come across a melanoma in a report and I will add it to my melanoma collection. Then perhaps once a month, I can put aside ten digital cases and use them for a teaching seminar. These sorts of things are much easier to do with a digital platform.
JT – We’ve talked about the achievement of KPIs being a driver of digitization in the NHS but there are also a number of barriers, cost for example.
That’s right and sometimes when we go into these departments, they’ll point to a scanner they bought five years ago which is still in its wrapping. The barriers are quite considerable and often the Trusts are not entirely aware of those when they start. I think it’s becoming slightly better understood now, but the investments in firewalls, storage costs, and IT connectivity are considerable. That’s one thing I do explain to these NHS Trusts, that you can go down this digital route, but it is going to cost you a lot of money.
When I was in South Africa recently for a conference there was interest there to use Source LDPath and I was approached by different clients who wanted to discuss going digital. The analogy I used was to say, “when you set up your laboratory, you didn’t get in a metal worker, a glass worker and a technician, to build a microscope for you. You bought one. So why are you spending all this money getting all these things together when we’ve got an off-the-shelf solution for you.” And that’s one of the messages I would give to NHS Trusts around the country.
JT – An off the shelf solution at a fraction of the cost.
AR – Exactly.
JT – Well let’s talk about the other emerging technology at the moment, the use of image analysis and AI in clinical pathology. Does Source LDPath plan to offer image analysis tools to assist analysis or diagnosis?
AR – Yes, it’s huge and very much part of our strategic goals and ongoing strategy.
JT – Using your own applications or from a third party?
AR – We don’t have our own at the moment, but we are developing. We are interested in implementing existing algorithms because if there’s a very good algorithm out there, there’s no need to reinvent it. One of the issues with AI is that you have to interface with the AI software. At least two of the AI companies we’ve spoken to have a universal platform. Once we link with those, we could access any of the existing AI algorithms through it, but we are also building our own system for that. We want to be able to link to any AI company’s algorithm and bring them in as we need them.
AI is astonishingly good. I actually lost a bet on this with a friend who’s not a pathologist. Some years ago, he said, “AI will be coming in and doing some of your job for you” He was right, and it’s now astonishing what AI can do, what I’ve seen it can do. For example, we use Ibex’s prostate algorithm through Source LDPath and we audited a series of cases from a hospital and corroborated the findings with two uropathologists. These were pathologists who do specialist uropathology work in a couple of teaching hospitals. It’s not that they are bad pathologists in any way, but each of them missed a cancer that the AI picked up, two different cancers in fact. At first, I wondered whether the pathologist would look at what the AI had picked up and say, “no, that’s, a common mistake, it’s not a cancer”. They didn’t, they said, “my God, it’s right.”
JT –So this means that AI will have a future impact on the role of the pathologist.
AR – Yeah. It’s phenomenal. And people are also saying that we won’t have doctors in ten- or twenty-years’ time. Well, I doubt that, not because I’m a Luddite, but perhaps we will need fewer doctors and their role will change.
JT – Well I’ll take that bet. I think there will still be doctors in twenty years’ time.
AR – Yes and personally, I am happy to have a message from an algorithm pop up as I’m about to authorize a case and say, “Alistair, do you know you missed a cancer?” I really want to know that.
JT – The analogy I would use is that of still having pilots in the cockpit of an aircraft when a commercial airliner can now take off, fly, and land using its computer systems. We still have pilots in there because, if anything goes wrong, they have human judgment and the ability to change their minds and to think creatively.
AR – Absolutely, and it’s easy for the AI to present potential cancer to you for your agreement or rejection which makes it ideal also for triage. One of the things we can do with the Ibex algorithm, if you have, say, a backlog of 200 prostate biopsies, is to offer to process them overnight with the AI. We can then say, “look at these four first because the AI says they’ve got cancer.” In fact, it’s not just identifying cancer, it’s doing all sorts of measurements. It’s picking up individual things that are real screening issues in pathology, things which are difficult, time-consuming, labour-intensive, and easy to miss.
AI will also be useful in quality assurance and some firms are already addressing this. We’re working on a quality assurance algorithm ourselves which isn’t on the market yet. So, it’s a very exciting future.
Going back to your previous question about whether we would use microscopes in the future, I think what will happen eventually, is that there will be a system whereby an H&E slide is automatically generated and put into a scanner. At the moment, you still have to put it in a scanner, but I think that process will become seamless.
JT – That’s interesting, because this is also a labor-intensive part of the histopathology process, creating the slide, cutting, staining etc.
AR – And if bots can do prostate surgery, I’m sure they can make a section one day.
JT – So to sum up, it sounds as though you are working hard to position Source LDPath for this future world that we are discussing. A world that incorporates digital scanning, LIMS, AI, and an entirely integrated workflow.
AR – That’s exactly it and looking back I would say that in the last four or five years, our ambition has been to completely redesign the way pathology has been done in the UK. We need a revolution in IT in the NHS because the NHS has had some challenges with its IT historically especially when trying to implement a top-down approach. My argument is that we should be building IT organically from the ground up and solving problems along the way. That’s our approach and that’s what we are doing with pathology.
JT – It’s wonderful to hear that Source LDPath is going to be a big part of this new revolution. Alistair, thank you very much for your time.
AR – You’re welcome.
