How Digital Can Transform the Pathology Laboratory

Chris Wolski:

Welcome to Clinical Lab Chat, part of the MEDQOR Podcast Network. I’m Chris Wolski, director of Business Intelligence for CLP, and today I’ll be speaking with Prasanth Perugupalli about the difficulties of implementing a digital pathology workflow, how to overcome resistance and why digital pathology is necessary for a laboratory success. Prasanth is the Chief Product Officer of Pramana, having more than two decades of experience and design applications engineering, go-to market planning and product cycle management, and is a big believer in the advantages of hardware, software, co-design. His vision is to bring a friction free transformation of pathology labs into the digital era.

So based on this introduction, I think I’m talking to the right person about the importance of digital pathology. So Prasanth, welcome to Clinical Lab Chat and let’s dive right in and get down to the importance of digital pathology. So I’ve been covering the medical field for probably about as long as you’ve been working in it. And I’m always struck that we’re still talking about the need to implement digital platforms. So I mean, what’s the holdup? Why are we still talking about why we need to do this, why this is a good idea, etc. What’s going on here?

Prasanth Perugupalli:

Thanks so much, Chris. I think if we would have a wonderful discussion around this at this point. When it comes specifically to talking about digital pathology and if we were to focus on why is it we are still talking about yay or nay from an adoption perspective, I think it is to do with the element of what exactly are the stakeholders expecting out of a digital transformation. So if you were to look at what happened in the radiology space about 15 years ago, almost the digital transformation happened extremely rapidly and seamlessly because the x-rays at the time the film got replaced and it went into the digital realm and people were able to go about doing their job just as if they were doing things with x-rays. There was actually a changeover, there was a changeover from the film to digital.

So there was a replacement. If you look at digital transformation across the board in any of the other lab practices with respect to how data is handled and how any of the data that comes out of, for instance, a flow cytometer has to be looked at in the digital realm because there is no other way to look at it. There’s been a very seamless adoption of digital in many of the labs in general. Digital pathology is a little tricky in that regard. At the end of the day, what we talk about is, of course, a workflow transformation with digital. Which means that you’re going to be doing a lot more things sitting at a computer screen or maybe on a tablet and much less dependence on paper. However, one of the key elements in the entire workflow, this aspect of sample under a microscope, the sample as we know it, the tissue sample sitting on a glass slide will still have to be made in the analog domain.

So, technically what we talk about when we talk about digital pathology specifically is essentially building a digital paradigm around the whole aspect of the tissue. The tissue itself remains analog, at least as far as we know today. There are technologies that I hope will come through that will also help bring some more interesting options in the future. So digital pathology hasn’t been a no-brainer for the labs because there was no replacement of the analog element, one of the analog pieces as it stands up. Now, of course, in the context of that, what ends up happening is there has to be enough other use cases, enough other efficiencies that have to evolve for lamps to be able to say we can justify the investments as well as the retraining and whatever changes we would have to go through to go digital in spite of knowing that there is a key analog block in the pipeline. So that’s where I feel it has taken longer than usual for this part of the lab to go digital.

Chris Wolski:

Yeah, that’s really interesting because you brought up radiology, that’s where I started 20 years ago covering that transformation, and it’s really interesting that analog transformation that you’re talking about. So one thing we talked about in the pre-interview was that fundamentally there’s often a disconnect between the digital pathology platform, which we’ve kind of alluded to and the labs established workflow. So is it that disconnect and really not the budgetary constraints because that’s always a thing that comes up. It costs too much to make the transition, the analog is good enough, paper is good enough, etc. Is that the real reason why a lot of labs aren’t implementing a digital platform? It’s more that technological disconnect that you kind of alluded to.

Prasanth Perugupalli:

I think it’s a combination of various factors. If we were to break the problem down into a couple of different quadrants, the budget element is always a challenge for sure. Especially like I said, when you’re not quite saving money by not having to make a glass slide very minimal element. If there was actually a replacement there, it would’ve probably been a no-brainer. However, we also need to realize that first generation systems, just like first generation, any technology is always going to be a lot more expensive because people are still trying to figure out exactly where to fit in the technical pieces together such that you can get economics out of it. And I think that’s a natural tendency and a natural phenomenon that is bound to play out in the early days of adoption of any technology. So I honestly don’t worry so much about the budget elements purely because we have been able to see ourselves that you can actually do digital.

When you look at the element of what are all the costs involved, there is of course, the cost of making the images. So like we said, the slides have to be made. That doesn’t change. So the cost of making images, the cost of storing images, the cost of preparing the images if you may, for downstream usage. So there are three different costs and storage as we know gets better and better in terms of pricing as time evolves. There are so many other applications in the world that are producing so much data that there is a roadmap in the semiconductor world to get storage to where it needs to over a period of time. There will be better and better technology advancements that come in to say how do we optimally store data? What kind of data to store? What is relevant, and how do you optimize that? This has its own legs in the technology world, there’s a lot of work that happens at very many institutions looking at how just to optimize storage of, it could be videos, it could be images, or it could even be a combination of those.

The aspect of cost of scanners, this is again, in my opinion, it’s a first generation problem. Every early generation device always tries to be sure that nothing breaks or it is built to perfection such that when someone is trying to go out there and evangelize a brand-new technology, the last thing you want to have is this aspect of saying that I have limitations. So everything is normally built to be the best of what you can, and this is something that is generally over a period of time you would realize that further research goes into the aspect of saying where are the places where I can back off a little bit in terms of whether it is the… I mean, performance will always be key, let’s say resolution to the images and image resolution, image quality is probably something that no one would want to compromise on. But the aspects of saying, do I really have to have the highest end customized robotics inside my system to deal with moving a glass slide that is a few grams in weight to be able to locate it in the right place for getting the work done?

Do I need to have extremely high-end mechanics that will ensure that there is absolutely zero vibration coming from anywhere in the building onto my system? These are various things that will continuously get, they continue to improve generation or generation. So today we are able to get to a point with Pramana, where we are able to tell people that you could digitize your slides at a fraction of the cost of what has been out there as an industry norm even until about a year or a year and a half ago.

Chris Wolski:

But it really is though in many cases. And what we were talking about in our pre-interview was a lot of what the issue, and we’ll talk about a little bit of how you’re solving some of this in a couple minutes, but it is that disconnect though between and get back to the radiology, the radiology example because I think both of us are familiar with that. I know one of the big complaints a lot of radiologists had back in the day during that transformation to digital when film actually was going away was that they wanted hanging protocols that looked like what they did in physical space.

And again, we talked a little bit about the disconnect with workflows that are being handed to an organization by whoever created the solution. Is that more of the issue? I mean, the budgetary of course, is important issue. But is that really more like you’re going to get more pushback from that laboratorian at the bench from the fact that you’re messing up their established workflow as opposed to they may not necessarily really be as concerned about the budgetary aspects of it, the rank and file. Whereas, a lab manager has to of course, talk about the budgetary side of it, but he also has to have laboratories who want to use this and have to be taken into account their work the workflow as well. So is it that disconnect? Is that what’s really the hold up to a certain extent?

Prasanth Perugupalli:

It is a big hold up as absolutely. So like I said, when we put the budgetary elements aside, assuming that technology will figure out ways to make the cost in line with what the market can bear, the bigger challenge has been the aspect of the integration. You talked about systems level integration into the workflows. In this particular case, we generally talk about the imaging management systems and the image management systems, the text management systems and such pertinent to the particular case. And how they do a two-way connection if you may, with the laboratory information systems. The laboratory information systems have been fairly archaic. They have to undergo changes on their side to be able to accommodate a whole new set of data types in very large images and such.

So what has been happening is that these two industries have almost tried to coexist with each other. So there are lots of solutions that are coming up that are finding the middle ground between what a laboratory information system looks like versus what is a digital pathology system producing and how to bridge that gap in between. There are several solutions out there. There’s no one cookie cutter solution if you may, the lack of standardization is a challenge there. So indeed, this is a space that hasn’t been fully explored. And what that does, Chris, is it essentially increases the burden of a lab that wants to go digital. So someone opens up an LIS on one of the computer screens and then, on another screen they have to open up IMS, an image management system. And another screen, they open up an algorithm that would probably run on one of the images and give an output. And why are they doing it that way is because like I said, the seamless integration hasn’t happened yet.

Chris Wolski:

It’s so interesting because this is just what happened with radiology. The same thing happened there as you probably remember and with other industries I’ve covered as well outside of the medicine, the same thing has happened. It really is so interesting that each industry has to go through these kinds of steps before you get this more integrated seamless workflow that I think we’re all hoping to get to.

Prasanth Perugupalli:

It almost seems like every industry has to learn everything all by itself.

Chris Wolski:

Again, it really does. It’s really strange and it’s also very frustrating. It’s very frustrating for everybody and particularly, I wouldn’t say frustrating, but I find that I have a lot of déjà vu moments. I’ve had a lot of déjà vu moments over the last decade or so because of that in covering different industries.

Prasanth Perugupalli:

However, having said that, I think we are seeing a tremendous amount of increase in the interest levels. I think COVID has helped a lot. It has opened up people to the notion of wanting to look at digital pathology, not necessarily as a panacea for all the problems that a lab has. But to look at bits and pieces of it, COVID has a load for labs to work with their pathologists and their consultants in a telepathology mode to get the job done. All of a sudden, however a bit of a pain it was with having to connect between several systems, things were getting done. And that is opening up, if you may, a new level of appreciation for wanting to go digital because it solves a problem. One of the biggest problems it is addressing today is a lab that is sending in any samples to a reference a lab, a larger lab at an academic medical center as an example for getting either some specialized testing done or getting a second opinion.

Today, the entire thing happens in a manual fashion. Someone sends in some slides, often they’re worried about sending in the blocks, the paraffin blocks because in case they get misplaced or something. So there is an opinion, the first opinion is created at the reference lab, then they might ask for the blocks to be sent in and then, the blocks will be processed for further testing and such. A lot of this is essentially painful both to the patient as well as the labs because there’s a lot of follow up, there is a lot of time lost mistakes happen and such. So these are the kind of things that digital is able to solve very quickly and it can be very contained in its nature. It’s a case that has been created, it has gone through its course and then, it has come to the end of life and it stops.

It’s been reported and it stops. And this is possible and you can imagine this happening anywhere in the world. So digital pathology is starting to happen a lot more rapidly now than what we had seen over the last, I think it’s the fastest ever uptake happening today than what it was ever in the last decade. And you can see it in conferences, in various shows that happen throughout the year that the chatter is happening, the chatter is high. People are realizing that there are some use cases that are more ripe than the others. And I think we will see a tremendous amount of activity in this space now.

Chris Wolski:

And I think you’ve kind of covered some of the big benefits. So let’s talk about what you do Pramana, and one of the things, my understanding is when we were discussing this prior, about a week or so ago when we were discussing this podcast, that what you really did is that you’ve kind of maybe, I don’t know solved is the right word. But maybe have at least, you’re addressing that design workflow disconnect, that you’re not sort of dictating what that lab workflow needs to look like. So once you go over how you’ve solved that and what does that sort of approach, and as you mentioned there’s no panacea, but what does that kind of approach give to a lab?

Prasanth Perugupalli:

Sure. So we are obviously not able to solve all the problems in the ecosystem. So there is a particular part of the challenge or the workflow optimization that has generally been not catered to very well. And this has to do with the glass slide itself, as I was mentioning before. So labs have to continue to make their glass slides, they have to continue to stain them and one of the challenges in the entire adoption curve has been this tremendous amount of new requirements that have been placed on labs in order to prepare to go digital with their glass slides. So there are a set of unsaid rules that have been formed over the last several years. A lot of publications talk about it and brochures of many of our competitors talk about it where they are essentially again, first generation systems, lots of constraints in terms of what has been built in into the systems.

So the labs have been essentially asked to make, do better with respect to their side of the operations. Making better slides, being very careful in terms of how the tissue is cut and how it is placed on a slide. When labels are placed, to make sure that they don’t stick out. The challenge with all of these is that it’s not that labs were deliberately making any bad slides in the past. However, when you go in and say going forward, you’re going to have to be extra careful. The immediate challenge that comes with that is actually of cost and resources. Technicians all of a sudden are a lot more vigilant, they need to be a lot more vigilant. There has to be some kind of quality assurance that is placed on top of some of the activities that probably did not need in the past.

So one of the challenges that we have heard whenever I have met with many of the lab managers has been this issue of how they have had to invest in other adjacent equipment sometimes in order to cater to the requirement for what are generally called digital ready slides. What we have done and we’ve demonstrated now that it is possible that you can utilize a lot of learnings from the image processing world, from the AI world to bring back the element of essentially dealing with however the slides are made today. How to make the most of it, and how to make the best possible reproduction of that glass slide in the digital form. So how do you do this? You essentially bring intelligence into the entire image making part itself.

In the image acquisition systems, we call them scanners. So intelligence scanners that are essentially able to do things in a fashion. I like to essentially call it fully autonomous and fully self-driving. Something that is able to not have to rely on a technician to have to make choices on whether or not to scan a slide, how to scan the slide, what kind of parameters to set, what kind of decisions to be made. These are all things that technicians are not generally meant. They’re not trained for this, this is a new technology. This is a new space that is still in some fashion, Greek and Latin for many of the technicians. So that’s the part that we have addressed in our first generation systems very well.

And today, what we are able to tell with a fair amount of conviction and with a lot of data to prove from some of our early mass deployments is that if a pathologist has been able to take a slide however it was made and go about using their microscope and make a decision diagnosis on the slide. With an intelligent system and associated software, we are able to pretty much replicate every bit of that information in the digital domain such that the pathologist can continue to do what they have done before.

Chris Wolski:

Right, okay. One of the things that you brought up earlier that I want to circle back to a little bit is, and you brought it again, you’re in a first generation technology, you talked about first generation technologies a little bit and certainly someone has to be the first to implement something. What would your advice be to someone who is looking at your solution? And I think the Mayo Clinic I think is using your solution right now in a big project. What’s the argument for an organization kind of touching back on that budgetary sort of issue? And sometimes the procurement could be a little bit of a conservative or part of the organization, they should be.

 But how do you make that argument to say, “Hey, we need to implement this” and it could be a first generation. We talked again in the pre-interview about some of your competitors that there’s some other solutions out there that have some good approaches to the digital workflow solution. How do you make that argument? What would be something, how would you recommend someone to a lab manager to make an argument for a new technology like yours or one of your competitors?

Prasanth Perugupalli:

So that’s an interesting set of questions there. Right, so let me tackle one at a time. So when it comes to the aspect of new generation or new technology, we call our technology today to be Gen two in digital pathology. And why is that gen two? When the last slides are prepared, they’re scanned. So a digital copy is made and then, there has to be some kind of an ombudsman that looks at the slides and decides whether they are good quality or do they have to be put back for a rescan. And then, also the aspect of saying is my data integrity okay, has everything been captured in the right fashion so that I can pass it along into my storage or cloud such that they can essentially be utilized for a pathologist or for an algorithm to do its job for the downstream.

First generation systems, rightly so, have essentially broken this down into individual parts and have addressed them independently. So they ask the labs to deal with the aspect of the specimen integrity, if you may, in terms of how the specimens are made and they are checked out whether they should be, they’re ready to be digitized. Scanners have essentially said, we do scanning and nothing else. The quality assurance has been left off to a set of people, again, histo technicians that are expected to open up one image at a time. Check for whether every part of the tissue, even tiny fragments of tissue sometimes get lost. Has everything been captured? Has it been captured properly? And is there a need for doing any, a follow-on capture one more time by putting them back in the scanner?

And then, so each of these, when they’re done separately, that’s what we have been calling as first generation and they’ve been doing it. The biggest problem to solve the was in the center, which was to say, can I make a good image? And that has been done fairly well. Gen two systems and what Pramana has been building off late are now systems that should be at a higher level, take away or essentially simulate a lot of these functions into one box such that there is no need, like I said, for the labs to have to do a lot of pre-work before actually a digitization process starts. There’s very little human interaction in terms of the actual making, the acquisition of the images and converting them into digital form, attaching them to other data from the patient and such. And the element of doing quality assurance, the burden of quality assurance on the lab, on technicians.

If that entire element can be brought into one box, then essentially, you have solved a big problem for the lab managers. And this can be done, again, it’s done in software, it’s done in systems, in hardware that is actually very heavily interlaced with software. And it has a roadmap to be able to make it much more economically than having individual parts. So this is something that we are able to convey to lab managers very well now. With Gen two systems, this is a huge burden that pretty much gets taken away. In one fashion when the equivalent of that is to say, you give us your slides and they run through our block our systems, and you get back fully quality assured, standardized data in your storage. It’ll appear magically and you do whatever you want to do with that data later.

So this is what we’ve been talking about and it also reduces the total cost of operation. If you could imagine the unsaid element of the cost of the number of resources along the entire chain is what we have essentially been able to solve. And the other element that we are trying to bring to the market and make some sort of generational change, if you may, in terms of how business is done, we believe that the scanning systems, the storage elements are continuously going to get more and more intelligent. There’ll be much more compute intensive and such as we go forward. And in the early generation of systems, this is having an electric car five or six years ago compared to today. You’re going to have a scenario where any CapEx investment that you would make to acquire these expensive systems today are probably not the right choices to make because two to three years from today, the systems could be looking much different than what they’re today.

And so, one of the things that we have been talking about as a new feature, if you may from Pramana, is the whole business model of digital pathology as a service. Wherein, we are discussing with our customers the ability to have them not have to have any upfront large CapEx investments, but look at this as a more of an operational expense maybe on a lease or a paper slide basis. While at the same time, giving them the assurance that the digital digitization part, the digital, the handling of the images and ensuring that the asset that is being created is future-proof as well as you can have confidence in what has been created. Those are the owners that we are taking upon ourselves.

Chris Wolski:

Okay, so the other thing you’re innovating on is just the way you’re getting the solution to the end user, in other words?

Prasanth Perugupalli:

That’s right.

Chris Wolski:

Okay. So certainly that helps with that argument that we don’t have a huge capital investment. It’s a software as a service type model that you’re following.

Prasanth Perugupalli:

And also, I like to tell people that it is in their best interest. It’s like essentially leasing a first generation electric car in some fashion. Why do you want to be stuck with something that is going to be so archaic?

Chris Wolski:

As the solution evolves or as their needs evolve, so the solution does as well, okay.

Prasanth Perugupalli:

Yes.

Chris Wolski:

And then, as you solve some more of these issues that we’re talking about with integration, etc, they’ll also get that benefit as well without having to make a whole new capital investment in other words.

Prasanth Perugupalli:

That’s great.

Chris Wolski:

Great. Well, let’s circle back then. I always like to circle back or look ahead and we’re going to do both here. So let’s circle back to our first question. So do you think in five years, you and I who have been around the block a couple of times will be still asking that question, why not digital, or will we finally have reached a time when analog systems are paper and all the things we’re just talking about are the rare exception rather than the rule? Do you think that you and I won’t have to have that discussion anymore?

Prasanth Perugupalli:

So who knows what the future has to hold and five years is a lot of time. But one thing for sure that I personally believe is that a good chunk of work that happens under an optical microscope, especially in the anatomic pathology lab, in the haem path lab, in cytology would have got converted to digital. Whether that means also that there is AI and there are automated decisions happening for sign-outs and such, I wouldn’t go there that fast. But the aspect of conversion from having to look under a microscope or to be in the lab at the glass slide with the box of slides that came with the particular case, as well as the paperwork that as came with it. That I can clearly see is ripe for getting transformed, whether it happens in two years or four years, it depends on at what point in the curve. Are you saying we have actually arrived?

Chris Wolski:

Right.

Prasanth Perugupalli:

Having said that, there are certain pieces to the puzzle that still remain. The one part that the industry still will have to figure out is the aspect of how to address the largest variety of things that happen in a pathology lab. There are certain specific use cases, certain specific subspecialties where it is actually very important to retain the complete data, if you may, in the three-dimensional space and of a particular location, of a particular feature because there is much more than what meets the eye in a plaster to the image format. So there is work to be done in the aspect of how do we store more data, how do we achieve more efficiencies such that you’re not blowing a hole in the bank in terms of the cost associated with going digital with a particular case versus what is the outcome that you got out of it. So we have some ground to cover, but I can clearly see that a big chunk of work would’ve gone digital from the perspective of just handling the case for sure.

Chris Wolski:

Okay. Well, great. Well with that we’ve come to the end of our time. Prasanth, thanks again for taking the time to speak with me today about this, I think critically important issue about workflow and leveraging resources. I think that we’ve covered a lot of ground here and I’m certain that there’s probably a lot more to cover. I also want to thank you the laboratory audience for listening. Look for more episodes of Clinical Lab Chat in the future and visit us online at clpmag.com and on all of the major social media platforms, and until next time.

Prasanth Perugupalli:

Thanks so much, Chris. It’s a pleasure to talk to you and great opportunity.