Aleksandra: [00:00:42] Welcome to the podcast my digital pathology trailblazers. Today in this episode, my guest is Prasanth Perugupalli he’s the Chief Product Officer of Pramana.
Pramana is a digital pathology company that is on a mission to revolutionize digital pathology but oh, are not all companies on the mission to do that? But Pramana has a unique aspect to this, and we’re gonna learn in which way Pramana is gonna revolutionize the digital pathology market. Welcome Prasanth how are you today?
Prasanth: I’m very well Aleks, how are you?
Aleksandra: I’m good. Thank you for joining us. We always start with the guest, you’re our guest of honor today. So tell us about yourself, tell us about your background, and then tell us a little bit about Pramana a little bit more than I said.
Prasanth: Thank you so much. I’m an electrical engineer by training. And in early part of my career I’ve been a designer, I’ve been a product manager mostly in the semiconductor world looking at early generations of cellular radios and such.
[00:01:42] So signal processing has been something that I’ve been close to and as a, some sort of a natural consequence, I started getting into image processing.
I work as a chief product officer for Pramana. We are about a year and a half old as a company and you talked about trailblazers and digital pathology. We are trying to bring a version of digital pathology that is more for the masses. Trying to make it in a fashion where, it is not necessarily a trailblazing technology anymore, but what it takes to be able to put it in the market, in the labs and in the hands of people that want to use it and be able to do it confidently.
Aleksandra: So I call my people, those who listen to this podcast, trailblazers, but also my guests are trailblazers to me. I call us all trailblazers because this technology is, has to wedge itself into their healthcare system. Everyone who’s making this happen to me is my fellow trailblazers.
Prasanth: Yeah.
Aleksandra: If you’re, and especially if you’re trying to bring [00:02:42] it to the so yeah, let’s go into pramana. How did this happen? And let’s maybe talk about it in conjunction with how did you move from being an electrical engineer to electrical engineer, working on images and then on images in the digital pathology?
Prasanth: Little bit of a history I was in the US working with large corporations in the wireless space for about 15 years. And at some point we had decided to move back to India. I had the unique opportunity to work with a research organization that was putting together some early days of looking at how we could utilize images in the medical space. This was a organization from Europe that I was heading. And I had a very I would say pure by stroke of luck, I think I met with a bunch of guys at the time who had come from LG and GE and a few other such organizations, all of us had one zeal in mind which was to say, could we build some kind of a system that has software, that has pieces of [00:03:42] optics and hardware in.
We built our startup back in 2016 a company called Spectral Insights that was built out of Bangalore in India. We were very clear about essentially utilizing the aspects of software computation and math in conjunction with the hardware element.
To say, could we build a true software, hardware, co-design, if you may which allows us to be able to solve a problem across any part of the stack rather than being pigeonholed as I would only build hardware, I would only build software. This was a journey that we started in 2016 and in 2019, just about when Covid was was hitting us nference, which is a company based sort of Cambridge, Massachusetts, a data science company. Which works in the realm of text and data analytics and is accumulating solutions that allow you to do a lot of multimodal data analytics.
Nference was looking for a partner to add imaging to its stable. Nference acquired [00:04:42] us.
Two years down as we were building these systems further and going into our early customers and such, we realized that co-design between, software and hardware was gaining a lot of momentum. Customers we approached were starting to realize the value.
Aleksandra: Showing a smartphone here for those who are watching on YouTube, that’s like the ultimate thing of software and the hardware coexistence.
Prasanth: Exactly. And so we were able to actually address a lot of the problems that the industry has been talking about for 10 years.
And we got a tremendous amount of encouragement from the trailblazers, some of our very early research collaborators at Mayo Clinic and such. And we realized that this technology is actually getting very ready now for prime time.
So Pramana was formed last year as a spinout of nference and we went on to independently build an organization with its own sales force and within with its own legs such that we are able to bring the digital pathology technology that we’ve been building to the markets.
Aleksandra: [00:05:42] So couple of aspects that are unique already in your journey is that hardware, software, merger, so to say. I’m showing the iPhone because in like real world and the internet of things is basically taking over stuff. But in digital pathology, like you mentioned, there is software companies, there are hardware companies, so there is the need for merger.
Prasanth: Yeah.
Aleksandra: So you have that and you have the parent company who’s a data company. Now it’s not just images before it was, okay, a pathologist looking at an image and there is data.
That’s a report, but how do you like extract anything quantifiable from the report? Now with all the algorithms with image analysis being done on whole slide images, there is a bunch of data. So this is like a triple merger software, hardware, data science.
Prasanth: Absolutely. And it’s fascinating when you start looking at it in that realm, What you realize is that there are opportunities just purely learning from [00:06:42] reading growth section notes to figure out what may have been the thought process behind that in terms of the slides that were created, the blocks that were cut.
The aspect of being able to merge text and text analytics with what happens with your hardware to be able to take advantage of inputs that can feed in directly to the acquisition system.
And the ability to say during that process, are there things I can learn in line that then can be utilized to be able to enrich the data further? When you start doing that in a fashion, just the way you describe the iPhone you realize that as you use an iPhone today for making images. The iPhone even lets you actually make some choices to say what object do I want to focus on eventually, is very, you could imagine there is a analogy that could be happily built in the pathology world and that’s the beauty of the software [00:07:42] hardware co-design.
And we are, I think one of the earliest guys who are. Trying to harness that. And a bit of that comes from I think, my own past history and background where I’ve had the opportunity to work with some very cutting edge wifi technologies and such. And we were looking at how a particular new wifi chip could get embedded into a brand new MacBook.
And, what are all the things that it can do to the user?
So this is a similar pattern in which we’ve been going about doing what we do and it’s starting to show some interesting outcomes at this point.
Aleksandra: Totally, definitely, I have to laugh always because I show a smartphone and you say iPhone, I know you’re from North America, because in Europe not so many people have iPhones. So I have to smile about that. But now, In all seriousness, back to our topic.
So you have like this holistic approach you have the technology, you have the data data science expertise. How did it happen that you ended up [00:08:42] in digital pathology? Did you go through a list of, I don’t know, relevant areas where you would wanna apply it? How did you end up in digital pathology? Why digital pathology?
Prasanth: That’s a interesting journey for us, Aleks. When we were part of this research organization especially around spectral imaging and hyperspectral imaging, we were trying to look at what are all the various kind of things we could do with hyperspectral imaging.
We had looked at fundus imaging as an example, and we had looked at imaging, leaves by flying small cameras from a, from a little bit of a elevation and we were doing foot imaging and such and interesting enough one of the problems that we had come across was looking at blood smears and detecting malaria parasite.
And actually hyperspectral imaging was a very good candidate to be able to pick up small traces of Falciparum
Aleksandra: I have to ask a layman question, tell me about hyperspectral imaging, what’s the principle of hyper spectral imaging?
Prasanth: Essentially with hyperspectral imaging, unlike in the brightfield imaging, we are [00:09:42] looking at the aspect of shining white band light, but then picking up data at various wavelengths through the use of filters and such.
That you’re not only picking up the notion of color, but also the aspect of what the behavior of the specimen is the actual content of the specimen towards various waves of light. And it opens up a whole new space that I think is imminent to happen in digital pathology as well at some point.
So someone had introduced us to the problem of looking at blood smears and detecting malaria. But then, we also realized that the industry worldwide wasn’t yet adopting digital pathology digital images by themselves for their routine work.
So adding the aspect of spectral imaging to brightfield would’ve made it a lot more of a challenge to be able to convince anyone in the field and we quickly realized that we had to go one step at a time.
There were cases where we had come across in India at the time where[00:10:42] organizations had spent a lot of money to buy these expensive whole right imaging scanners, and they were collecting dust because either people didn’t know how to use it or the requirements that were put on the labs to say, thou do so much, so many different things in order for the scanner to work were just daunting.
And what originally started out saying, we wanna solve a problem in malaria, became a mission to say. Looks like the biggest opportunity is to actually build a a very heavily software driven system that takes away the enigma of, how I would go digital in a lab?
Aleksandra: what was The biggest surprise or like the biggest, like you did not expect this in pathology lab when you were in there, what was the surprise for you?
Prasanth: I would say.
Aleksandra: Or something, Ha why is it done like that?
Prasanth: I know I had a immense amount of respect for doctors growing up.
In India you would either be an engineer or a doctor. And those who cannot be doctors became engineers, the biggest surprise to me was [00:11:42] the fact that there were about, in some cases, 12, 15, 20 slides that someone was going through, but it all resulted in one report.
And it told me that somehow in the mind there is a memory element that was being built up to be able to read through all of these and continuously go about building a case, and being able to write a report. And that was to me more than a surprise to my respect for pathologists went up so much when I saw that.
Of course the immediate question was then why wouldn’t you use digital? And it appeared to me at the time that it should be possible to use digital and put all of these on the same screen, on a very large screen and look at them. And then we realized that the labs were struggling to be able to actually go digital at all for a variety of reasons.
One of the biggest things was just the price points but more importantly, I think at the time I had realized that it would be so difficult for them to change their processes or they [00:12:42] have to adhere to so many different things in order for getting good images.
It felt that there wasn’t, the opportunity to say if I brought in a bunch of image analytics or image processing onto the place where I’m making image. A lot of the problems can be solved in a much cleaner manner.
So that was our impetus, to get going and we see that today, everything that Pramana is talking about today is all about saying, how do you liberate the technician, the lab manager, and the pathologist from having to get retrained or look at what kind of compromises they have to make either in terms of their own work style workload are compromises in terms of what they’re offered on a screen to make a decision.
That’s what Pramana is trying to achieve to address today, the aspect of making it easier, if you may, for anyone in the ecosystem to say I can achieve digital, [00:13:42] both from a technology perspective as well as from a commercial perspective the cost of getting there and such.
And in the next generation, we are now looking at how much more enriched data can be generated from these systems, which again, is a second, is a next generation element to digital.
Aleksandra: So I have to, I was sighing when you said that those scanners were collecting dust, I have this vivid picture in my mind.
I am, my PhD was funded by a scholarship and with the scholarship, I went to Ethiopia wants to visit a pathology department and there was my co-scholar we all went to for the scholarship to Germany and he just finished his scholarship. He got his PhD. There was some kind of grant that bought all the lab equipment, PCR equipment for this lab.
It was a fully equipped lab. He was there as the future lab director with all the know-how from Germany. Everything was collecting dust because the reagents were too [00:14:42] expensive for that hospital.
Prasanth: Absolutely.
Aleksandra: And it’s like when you said the scanners were collecting dust. I’m like, yes. And how much of this stuff is happening across the world, across science, across healthcare?
So I am super excited to hear how are you going to give us, give the market, give the digital pathology space the next generation? What are you guys offering at the moment? When somebody wants your thing, what is it? How can they benefit? And how is it done technologically different than it was in the first generation way of scanning and doing digital pathology?
So we’ve been
Prasanth: talking a lot of late, about a second generation of scanners and essentially, It is to say that scanners should become intelligent enough that they’re able to figure out what kind of a specimen is offered to it. What may be all the things that can generally happen in a lab in the due course of how work is [00:15:42] done.
Whether a cover slip is not able to be placed perfectly well or a label sticks out or for that matter a microtome is just a little bit out of calibration and you don’t get perfectly well made slides the aspect of seeing how scanners can be made more intelligent and more aware through the use of algorithms, through the use of sensor.
And take away the burden from the technician as an example to say, should I load this sample in this scanner? Should I not load it? Should I have to make some choices in terms of whether it is of focus, sample points or these stacks and various other elements but the ability to say, why couldn’t you build a fault tolerance system that is able to detect things, while the slide is still inside the scanner?
And with a little bit of a differentiated parameter set, just like how a, you would expect a technician to do that. Can technically be learned by the scanner and the element of essentially building a capability [00:16:42] that takes away the human dependence is what we’ve been talking about as our second generation digital pathology systems.
This, we think is the second generation for the industry what we are offering today is essentially a full fledged solution that allows us to be able to bring in our scanning systems, our software and deploy it at customer premises. And essentially be able to give confidence to the customers that they could literally hand us their slides and we would essentially give them back assured, fully quality tag images.
In a standardized format and back into their storage space, how are we able to achieve that? Is through the aspect of we don’t have armies of people running these. In fact we have a lot fewer people than what a customer would normally expect to have.
While, they go digital, which by the way is a area that is heavily discounted in general. Folks understand only after they bought the scanners and want to go digital, how much of a human intervention is [00:17:42] needed. Yes.
Aleksandra: Everybody’s talking about, oh, pathologist and the image and it’s equivalent or non-inferior, and I have not yet seen a paper maybe I should read more papers, which is my resolution for next year. But something focusing on the previous work, the pre viewing workflow. Pre-evaluation workflow, and you even mentioned crossing like yeah.
I have heard this there, here and there, but not really, this is not really a fullfledge discussion yet how to start there and maybe when we are already there, there is another step ahead that we can I’m not even talking about all the pre-analytical and standardization of fixation and all this stuff, like that’s a topic for a different discussion. But yes, definitely this is a heavy burden on the staff operating those machines and.
Prasanth: We have addressed that through building software the current flagship program that Pramana is involved with.
[00:18:42] Which is in my opinion the single largest digitization program that is happening at Mayo Clinic in Rochester right now. Almost 10,000 slides of scans happening every day and being really manned by two.
And so the element of being able to achieve a highly efficient scanning operation, which includes all the way from essentially saying, I take the slides as they come through, no pre-assessment of what is on the slide, no cleaning of the slides, no assessment to say, is the label sticking out?
And things like those all the way here is the quality tag that says this much percentage of fields are potentially having a soft focus on the slide. And all of this being done completely in an autonomous manner is what we’ve been able to build and deployed and run it now for several months.
And so we have now I think, proven to ourselves and to our [00:19:42] customers that this is actually
Aleksandra: so you are running a massive archiving project at Mayo Clinic and I recently had the guest from Mayo Clinic, Dr. Rish Pai was also involved in the digital pathology initiative there. Okay. So let’s talk a little bit from the business perspective, service versus product.
What you’re offering now is you have your product place at the customer site, and you’re offering this service or as this archiving service. Why that? And is that gonna be the model? How is this benefiting the clients?
Prasanth: That’s a very interesting question. The reason to offer service as an pathology as a service, which results in customers not having to worry on their end their resources and training and retraining and complying.
Aleksandra: Just wanna emphasize something because digitization of slides is a service you can send it to somewhere and they will scan it for you. Your service is different, you [00:20:42] come in with your machines which for YouTube I have to include clear how this scanner looks like it’s a different looking scanner.
It’s whenever you guys look at it, you will not think it is a scanner. For the longest time when we were talking with the, with you and with the team, I was like, you, they’re probably like different objectives within a box it’s nothing like a box, but yeah. So the machine is placed at the customer and your people are doing their job at the customer site, so there’s no disruption to whatever is happening in the hospital and that’s different.
Prasanth: The aspect of customers, whoever we have talked to in the early phase I, we felt the biggest impediment for them to go and do something at scale was just the fact that their early trials and their early experiments that they have done pilots and such, resulted in a situation where I don’t think anyone had the confidence that they could do things at scale.
They were worried about all the issues that [00:21:42] they ran into and so when we went to our customers, early customers and made an offer to them saying we would take it on our chin and essentially literally have you give your slides and then we’ll give you back digital outputs that are fully QA assured that are DICOM compliant and such. It was more to actually lower the barrier for the customers than anything else. It was to give them assurance that we are able to do that and also to be able to do that at a very commercially viable price point.
Aleksandra: Especially if somebody already tried, invested and then justifying, like trying out again for the same amount of money is, there’s gonna be a big pushback.
Prasanth: So something had to be different, so we had to innovate and I think one of the innovations, which is gaining a lot of interest across the US is this aspect of saying that customers don’t have to make capital commitments.
Upfront to buy these systems and also [00:22:42] have to decide on how many times a year would I actually get a service call. My way to look at it, Alex, is we are in the earliest phases of digital scanners. They are going to dramatically change over the next couple of years, so the ability to convince the customers that they shouldn’t have to make these calls and actually, buy these systems that will essent very soon become bricks over the next two to three years, if you may was actually something that we realized was a real need to address.
And so our vision was right on to say, how about we demonstrate our confidence in being able to do the digitization the way we are claiming it can be done, by essentially walking the talk and saying, we can do this on our capex and you pay on a pay per slide basis and it is starting to work beautifully.
One of the biggest challenges that we had to deal with, I think more than just the [00:23:42] aspects of operations and how we thought it would run, is just a variety of specimens that come your way.
your. Interesting enough, the variety is less about the tissue, but more about the aspects to deal with the labels and the IT system for a particular year being different from some others. And sometimes ink has flown over a 2D barcode and hence you cannot decode it anymore. There are lots of interesting challenges.
Now the business model that we’ve been able to put out there is essentially to take away all that friction. we are able to go back and tell our customers that we can continuously improvise and adapt to newer challenges as they come with newer sets of slides from different departments or sometimes it is slides that are coming in from other locations for second opinion and which have a whole different type of syntax, if you may, and how the slides are made and how they’re handled.
We’ve been able to demonstrate that the intelligent systems concept is able to, be [00:24:42] resilient enough that it can actually adapt very well. And customers don’t have to change their budgets.
They don’t have to, they’re not caught by a surprise. They are aware that, the cost of going digital for this many slides in a year is going to be about this much, and they don’t have to worry about maintenance contracts and how long before, a service professional shows up.
Most importantly, we are offering the ability to, monitor and handle any kind of challenges that the customer might face. The part about, essentially taking away the pain of making images from the customer and using a wide network that could be worldwide in nature.
That is taking learnings from various locations and being able to address day to day challenges that will happen on the field. That is where I feel the industry has to move. The end point is not scanning the end point is way downstream and the aspect of making images, [00:25:42] handling images, enriching the images with whatever data you can, should almost become a table stake element.
Aleksandra: And you say this you said this sentence, the endpoint is not scanning when the endpoint is scanning? People don’t see the value of digital pathology.
Prasanth: Exactly.
Aleksandra: Because, okay, if we have a pathologist we can do the same on the image, on the microscope, faster without system, without upload, download. What’s the differentiator if these, the digital and the glass are non-inferior?
Prasanth: Yes.
Aleksandra: That means glass is non-inferior or the same so let’s keep doing glass.
Prasanth: Yes.
Aleksandra: And so far the added value was image analysis. Image analysis in terms of quantification in the simple form of IHC quantification. Then deep learning came to the scene and prediction models were being built. So this is still like a huge added value.
Prasanth: Absolutely.
Aleksandra: But there is a lot of stuff that is not being addressed, like [00:26:42] tagging, like the QC and like everything that like in between, that happens between the scanning and all this fancy insights that you can get with your fancy algorithms or less fancy whatever. I love image analysis, that’s my background.
So anyway, but yeah, this is definitely something different, not addressed. I wanna ask you because when you’re saying about this, we are here with the scanners in this point, you’re farther, there’s gonna be a transition. There’s gonna be a transition of institutions that invested heavily in the scanners are still using them and leveraging as much as they can leverage it.
But then they’re seeing this next thing and that is better and then on one hand, how do you justify abandoning all this investment? On the other hand, how do you justify continuing doing something that’s less efficient that it could be, what are your thoughts on that?
Prasanth: That’s.
Aleksandra: And with the phone, it’s [00:27:42] something everybody can afford more or less. You just ditch the other phone and take the new one. It breaks the glass, breaks you buy a new phone now.
Prasanth: Yep.
Aleksandra: Let me know what you think about that.
Prasanth: It’s very interesting the way you describe the phone of course, scanners are much more expensive. However, at some point technology is going to have to drive the next generation decisions because when labs and hospitals and institutions are talking about digital today, they are talking about holding the digital images for a while.
Algorithms are being built not for today alone, but for going well into the future. And one of the challenges that you always have with earlier option technologies, is that, things work to a certain point with a lot of compromises and then there’s gotta be some inflection point somewhere where you say, everything I do needs to be future proved.
And I have to make a choice and I’ll give you a perfect example millions of slides have been scanned, and they [00:28:42] don’t know what is the quality of the data that is actually in their storage because it has been too expensive to go back and open them up and check to say, what do we do with these?
Do they have to be re-scanned and or even to go in and say, this is an area that I should not have used for my algorithm. So at some point you have to make that choice to say what is, it’s a yes, it is a, some cost versus, how long before all this data would start to become not so attractive anymore because technology is changing.
You have to make a choice at some point, you have to look at the aspect of what is the new technology bringing. When you start to get to that point where you’re able to make such calls, there is new data getting added. So when with new data, algorithms will evolve and I think they will go at higher level, the challenges that the algorithms have to solve will change too.
The thing is when pathologists have been able to use glass lights in whatever form they got to make their [00:29:42] decisions. If algorithms are getting constrained in terms of saying we cannot achieve that, you have to realize that technology will continuously evolve towards that new thing.
And customers have to be very of that and that is indeed why when I talk to a lot of customers, I’m openly telling them, I think you should stop buying scanners anymore because technology is gonna change for a long time things were at a very slow pace, right?
But now you see digital taking off in a fairly large fashion, and a lot of it is driven by algorithms, like you said, and just the aspect of having a whole new appreciation for tele pathology and various other elements. But as it is catching more tailwind, technology is going to evolve faster and as it evolves faster, you have to continuously keep moving.
Gone are the days when you keep your phone anymore for five years. So it’s the same situation that you are going to move into and what does that do? [00:30:42] It results in new business models. And that’s indeed what we are talking about which is please don’t pay for this.
Aleksandra: That is basically the definition of innovation, you move with the world and like when you said, okay, there’s not really, there is not a way to keep the old stuff.
And I thought when you were saying this, those old scanner, I reminded myself, I have the picture, how many cameras? I have I don’t know, six.
Prasanth: Yeah.
Aleksandra: Include maybe eight because including two non-digital cameras that I have.
Prasanth: So we see that and one of the things we talk about many technology perspective as an inflection point I think from a intelligent scanning and intelligent, I would say enrichment data enrich is what I think we pioneered.
I think we were the brave guys to go out there and look at how to do this right? Which is to say, how do we do volumetric scanning? Scanning in a similar fashion as how any pathologist would use their microscope in a fine focus mode.
And the, that actually changes the paradigm quite, so it [00:31:42] is truly a big change, a step function. So it is good enough of a reason in my opinion and I don’t want to sound like I’m a sales guy trying to sell Pramana scanner, but I think
Aleksandra: it’s your project. You believe in what you guys are doing, so totally go ahead and be a hundred percent behind what you guys have.
Prasanth: More so than anything else, it is actually very good for the industry when you are able to say that, all the impediments, all the problems that the industry has seen as people have tried to tinker with the early systems can actually be solved in a very nice fashion.
The second part of it is then also the appreciation to say on a daily basis there are going to be challenges with specimens in the lab. It is easy to say that a technician should be more careful and you can invest more money in a more advanced cover slipper and, have different ways of printing labels onto slides and things like that.
In the general scheme of things, when you talk [00:32:42] about beyond the early adoption elements and mainstream adoption you have to cater to the masses and to cater to the masses means that you have to solve a lot of problems as they occur in the labs. And one of the things that we’ve been able to do extremely well is this aspect of using robots in a fashion that haven’t been done before.
Aleksandra: I’m smiling cuz I have seen a video of your robot and it’s the most hilarious thing I hope we can put this on YouTube as well over this.
Prasanth: Automation and computation can change things, you’re going to see a lot of changes coming there as well from us, as well as from many of our competitors. So all in all, the aspect of when it comes to business models the element of saying this technology now is at a point.
It can be essentially treated like most of the other lab equipment, we also recognize that there are some customers that will need a different way of looking at things. So we are starting to build ecosystem partnerships and [00:33:42] distributor partnerships and such where we can offer things at various other strata as well.
Aleksandra: I’m really fascinated. I’m looking forward to hearing more about your technology being brought to the hospitals. And I know we’re gonna be talking again. So before we go today, before we finish, let us know where can we find more information about Pramana and the second question, which is the next conference you’re gonna be at so that people can talk to you and ask about that.
Prasanth: You can obviously visit our website, www.pramana.ai and you will find lot more information and some videos and such these days. I will be at the Global Engage, we do plan to be at the US Cap , in March with a fairly large presence with some new products that we are going to be demonstrating.
Some very interesting new visualization engines that that we are seeing a tremendous interest from customers for. So really looking forward to bringing some interesting things out there and showing them at the US CAP.
Aleksandra: Fantastic. I’m gonna link to your website [00:34:42] in the description and my resolution, my nearest resolution is go a little bit more on tour, on the conferences and actually play with all those things that you guys are telling me about on the podcast and touch them and talk to your people.
So I’m super excited about that. Thank you so much for being my guest and have a great day!
Prasanth: Thanks very much Aleks! You too have a nice day!
