[00:01:40] Aleksandra: Today, my guest is Michael Grunkin, the CEO of Visiopharm. Michael co-founded the company when digital pathology was just being born. And now 20 years later, Visiopharm’s software is used all over the world for quantitative pathology. Today, we will talk about his 20 years perspective of the field of digital pathology.
[00:02:03] Hi, Michael, welcome to the podcast. How are you today?
[00:02:07] Michael: Thank you very much for this opportunity. And let me just kick off by saying happy new year and thank you for the invitation.
[00:02:14] Aleksandra: Happy new year. This episode is an episode that is during the year of Visiopharm’s 20th anniversary. The first Aperio whole-slide scanner was released in the year 2000, and then two years later, Visiopharm was there. You are one of the first image analysis players on the digital pathology arena. And now after 20 years, you are still standing strong.
[00:02:42] Let’s talk about what happened in those years, and let’s start with how digital pathology looked 20 years ago. How was it back then when the first Aperio scanner came out, and how did you even have the idea to start this kind of company?
[00:02:58] Michael: That’s a funny question to get 20 years after and trying to look back at that. But if we really look back 20 years ago, it’s probably an overstatement that we had digital pathology in the context we think about it today. We certainly had the first scanners out, but it was really a very nascent technology. And we were actually focusing very much on looking at digital microscope images. We spent a whole lot of time integrating to different digital cameras and stages and linear encoders and developing auto focus and everything else.
[00:03:38] I think the concept of whole-slide images had not really become household. It certainly existed in the mind of Dr. Bacchus and Aperio, but it wasn’t really household at the time. But certainly a lot of the image analysis problems are still the same today as it was back then. It really revolved a lot around identification of cells and tissues, quantification of subcellular biomarker expression, maybe even co-localization of biomarker expression. Finding tumor regions, other structures and sub structures in tissue, just like it is today. But I would say that back then the use of digital or computational pathology, even in a clinical setting, was almost unthinkable because most of the technologies that are really required to accomplish that were not ready for primetime use in a clinical setting. And to be sure that regulatory guidelines around AI and computational pathology was not really available, those are actually quite recent you, of course, know.
[00:04:57] Aleksandra: Yeah, this is the last couple of years.
[00:05:00] Michael: And I would certainly say that, even today, going through a digital transformation is challenging. I’m sure we don’t need to talk much more about that. But 20 years ago, and for several years, most of the customers we saw and probably most of the institutions that would adopt aspects of digital pathology were biopharma. And of course later on, we also sold more and more in academia, academical medical centers joined in. But I think one of the major challenges in the research market for, let’s call it quantitative tissue pathology, was and is really what I would call hypercomplexity. And what I mean with that is basically the number of applications multiplied with the number of types of tissue, stains, imaging technologies, and research questions. That’s a very large number. And either you need to provide a solution capable of addressing this hyper complexity, or you need to focus on a subset of problems you want to solve. I think that was the dilemma back in the day. And focusing on a subset of problem is not as unproblematic as it sounds, because the questions that scientists ask keep evolving as a function of their own insights. So you really want to offer solutions that scientists can grow with, and not a solution that limits the laser research project. And that’s certainly a requirement from any strategic user of image analysis. And of course, 20 years ago, that was very much biopharma companies, and it actually is today also.
[00:06:50] But I would say that today image analysis has become a mainstream application for most tissue based research. But of course not as we know yet in routine diagnostics, but we are seeing some very, very promising signs.
[00:07:08] Aleksandra: started very shortly after the whole-slide scanning entered the market, but that was not your main modality. You were working on static images. You develop both on the imaging side and on the research side of image analysis. How did this evolve and how did artificial intelligence come into play since you started Visiopharm?
[00:07:34] Michael: Well, I would say very, very few labs had access to slides scanning 20 years ago. And a lot of biopharma companies and other research institutions were probably still skeptical of this, but they had the research microscopes already. But certainly, for both biopharma and certainly now pathology labs, it goes without saying robust, fast and high-quality slide scanning really is an enabler for the entire downstream computation pathology market, whether it’s in research or in diagnostics. I would say if we just for a minute focus on the research field, needs and expectations, they continue to evolve in many different dimensions. And I won’t really touch on all of them, but I would say one very, very significant and probably also recent development, especially in the wake of several breakthroughs in immuno-oncology, is this need to understand the tumor microenvironment. And there is a very strong and also growing demand, not just for multispectral and even hyperspectral imaging, but also for tools that are capable of analyzing spatially resolved and multidimensional data sets, and provide advanced data visualization tools on top of that. I think we’ll continue to see a lot of development in that particular field, primarily driven by the need to discover and develop new companion diagnostic tissue biomarkers as part of subtract diagnostic model, that is not only now a regulatory necessity, but it’s also proven quite successful. think in the wake of this, and even before that, there is both a demand for and expectation for robust automation of image analysis. And it’s almost, I would say, an understatement that AI and deep learning is a breakthrough and even a quantum leap, that’s on a big scale replacing traditional machine learning for things like image segmentation, image analysis. I think it’s made it possible to address a much, much broader class of image analysis applications. And it’s become possible to achieve a robust high-quality and very generalizable results that could not be achieved with classical machinery. So I would say it’s really changed the computational landscape and also the challenges a lot. And it’s made it a lot simpler to develop new applications, which can now be handled by the researchers themselves. they can develop these applications, just using a simple teach-by-example if they use AI.
[00:10:44] So AI has evolved image analysis, I would say, from a science, sometimes even an art form, to a craft that could be used by an application domain specialist. And that’s, I think, a new development. So the bottlenecks are changing as well. It’s becoming access to data, it’s becoming training of networks, it’s becoming compatibility with new imaging devices. I would say the ability to support the analysis of new and increasingly complex data sets.
[00:11:16] And of course, goes without saying, connectivity with existing digital infrastructure and certainly as computational pathology is beginning to make its way into routine diagnostics, we are seeing that regulatory requirements is something that will demand a lot of attention and investments if you want to make a difference to diagnostics.
[00:11:40] Aleksandra: Definitely. I hear two technologies that enabled breakthrough that everybody knows about, and everybody talks about, which would be the whole-slide scanning and artificial intelligence. But you mentioned two things that not everyone is focusing on, but they’re equally important. The interoperability of systems and the visualization of the huge amounts of data.
[00:12:08] I think keeping that in mind is so very important, and these are the two things that I don’t hear so often people say that’s helped advance the field and actually … this is all decade. It sounds also long ago but it’s just couple of years ago, AI, I remember being senior stand at the conference in 2019. This was, I think, then your AI model was new. Now it is mainstream, like you said.
[00:12:36] Michael: I totally agree.
[00:12:37] Aleksandra: Yeah, so it was slow, slow, slow, slow. And then a couple of years ago, boom, and this great increase of capacity. So what would be the standout achievement for Visiopharm for you? What are you most proud of for having accomplished in those 20 years?
[00:12:55] Michael: It’s funny to stand here and look back 20 years, and it’s almost difficult to pick just a few things. But given we have limited time here together, I chosen, I would say three things, that I feel stand out. And certainly on the commercial side, I’m very proud of the fact that we’ve been able to grow more or less organically from zero to 55 people and a significant revenue stream up until where we got our real first A Series investment.
[00:13:33] But until that time, we had really insisted on being in the trenches of creating value that our customers were willing to pay for. And that allowed us to establish a loyal customer base in more than 40 countries, and at the time with around 800 licenses being used on almost daily basis. And over the years, it was very clear that we have become [inaudible] the go-to solution providers for strategic uses of image analysis and computation.
[00:14:08] Aleksandra: Everybody knows Visiopharm. You say image analysis, Visiopharm is the name that always comes up.
[00:14:14] Michael: I would say certainly when people hit the boundaries of whatever solution they’re working in and need heavy tools to break the way through their problems, we have become that we had grown in the most recent years very fast, we started suffering growing pains. And I know that might sound like a luxury problem for a company, but I have to say that one shouldn’t underestimate neither the pain and nor, I would say, the dangers of having growing pains. So it was very important for us to go out there and raise our A Series round to grow the organization. And of course I’m very proud of the trust that our new investors showed us. And even more of course, that we’ve been able to deliver on their expectations and live up to that trust. And that almost brings me to the maybe most important point, because I’m very proud and also grateful for the team that we’ve been able to build at every level and every function of the company. And in fact, many of them have stayed with us for many years. And-
[00:15:40] Aleksandra: That was follow up question, do you have people who are … how long have the longest employed employee been at Visiopharm?
[00:15:48] Michael: actually I can say, with considerable pride, that we have two 20-years anniversaries coming up this year, and that’s beside my co-founder, Johan, and myself at this-
[00:16:00] Aleksandra: You have two other people from the original team?
[00:16:03] Michael: Absolutely.
[00:16:04] Aleksandra: This is fantastic.
[00:16:06] Michael: And I would say that it’s also, it’s really a privilege that there’s so many people. And I mean, today we are 120 people, and everyone is taking part, not just in pursuing a common vision, but on a daily basis develop that common vision. Which, briefly stated is that we want to be the enabler of precision pathology or precision medicine as practiced in tissue pathology, if you will. And I think we all very much believe that precision pathology is what it will take to make a meaningful contribution to the development of new treatments, to identification of responders. And ultimately of course, to patient outcomes.
[00:16:59] Aleksandra: Doing this for 20 years is really a great achievement, so congratulations on that.
[00:17:06] Michael: Thank you very much.
[00:17:08] Aleksandra: How are you celebrating this year?
[00:17:10] Michael: Well, unfortunately that has not played out as we had hoped. had naively thought that we could celebrate this with all of the Visiopharmers and our partners and friends of the house at the beginning of this year. But as I think it’s clear to everyone, that’s really not possible, feasible, even legal at this time. So we have to postpone our celebrations until we enter a warmer and sunnier time with much less disease than you’re seeing right now.
[00:17:41] Aleksandra: But the plan is still to celebrate as a live event with everyone that can join?
[00:17:47] Michael: Absolutely. And we can’t wait.
[00:17:50] Aleksandra: Well, then happy celebrations. And thanks so much for taking the time and telling us about this fantastic, inspiring journey.
[00:17:58] Michael: Thank you very much. And thank you for inviting me to this.
[00:18:02] Aleksandra: You’re welcome.
