1.4 — Ibex with Joseph Mossel

Transcript
Subscribe
All The Next Horizon Podcasts

When it comes to your health, nothing matters more than an accurate diagnosisIn this episode, Bill talks with Joseph Mossel, the CEO and co-founder of Ibex Medical Analytics, to learn more about the role that artificial intelligence plays in their pathology labs. Tune in to hear what this means for you and the future of healthcare.  

Read more about the Ibex Second Read System.

Guest List

  • Bill Pfeifer is a Dell Technologies messaging creator. He focuses on the emerging tech landscape and strives to ensure everyone is ready and excited for what’s coming next.
  • Joseph Mossel is the CEO and co-founder of Ibex Medical Analytics, the pioneer in applying computational pathology and AI to cancer diagnostics.

Bill Pfeifer:

Hello and welcome to The Next Horizon, a Dell Technologies podcast. I’m Bill Pfeifer and together we’ll be talking about emerging technologies, their potential to impact society and what you need to know today.

Our guest today is Joseph Mossel. Joseph is the CEO of an Israel based company called Ibex, which provides AI based pathology evaluations. We connected with Joseph through the Dell Technologies Capital team who were excited at the potential they saw in Ibex to both help society and succeed in their business and they provided some investment in the company to help Ibex accelerate their growth. Now, the team at Ibex has certainly been busy expanding their AI based work to help improve the speed and accuracy of pathology results for cancer diagnoses and Joseph was kind enough to take a few minutes out of his schedule to talk to us about how they’re using technology to make a very stressful situation a bit easier on people going through it.

Joseph, thank you so much for your time today and welcome to The Next Horizon. If you don’t mind, let’s start by talking for a moment about you and Ibex as a company. How did you get into the business of AI based pathology? What’s the overall passion that’s driving you to do this work?

Joseph Mossel:

Hi Bill, it’s a pleasure to be here. So the starting point for Ibex for me and my co-founder Chaim Linhart was that we wanted to do something which we felt that was meaningful, that has impact on people’s life. And very quickly we narrowed down on the healthcare field as an obvious choice for that and also considering our skillset, data sciences, machine learning, algorithms. Like most things in life, the insight to do specifically what we’re doing now came from a conversation. We talked with my brother-in-law who happens to be a pathologist and we learned a lot of interesting stuff about pathology and the discipline, but what we learned that really surprised us that pathology is practiced on a microscope and not on a computer screen. But what we also learned that this is changing now, that pathology is shifting from doing it on a microscope, looking at tissue to pathologist now looking at an image, a scanned image on a computer screen.

Now this gives the opportunity for people like us to start analyzing the data. Once we started getting into the field, we gained another insight is that there’s a real need for technology to move into this field. And the reason for that is that there’s a growing gap between the ability of pathologists to deal with the workflow because there are less and less people entering the field on the one hand. And on the other hand you have a growing incidence of cancer. So there’s this growing gap between the capacity of pathologists and the actual need.

And this is what happens if you look at the developed world. If you look at the countries like China or India they’ve never had the enough pathologists and it’s difficult to see how they bridge the gap. And this is where the technology can come in and help bridge this gap between the capacity and the need.

Bill Pfeifer:

So it’s a growing problem that’s going to be creeping up on us pretty fast. Now, as we were gearing up for this episode, I was using terms radiology and pathology as fairly interchangeable as well as detection and diagnosis. But I understand that’s not at all correct and there’s actually a coming crisis in the pathology practice. To help keep me accurate and to help our listeners understand two of the fundamental terms to this conversation, would you mind giving us a quick introduction to the terms and to pathology as a practice?

Joseph Mossel:

Let me take us quickly through the cancer diagnosis pathway. So you have a patient and they’re manifesting some symptoms, a pain, something that indicates that there might be cancer. Typically, the first step is that the physician wants to detect where the cancer is and what plays the major role is that as imaging technologies, radiology, this can be MRI, CT, mammography. Using this you can locate where there might be a tumor. Still having done that, you don’t know for sure that it’s cancer and you haven’t really diagnosed it. So once you detect where there is a possibility of a tumor, you take a biopsy, a tissue sample. This is sent to a pathology lab where a pathologist looks at the tissue at magnification and then they can reach a real diagnosis of whether this is cancer or not and also grade it. So this to a very large extent determines the treatment.

So this is the practice of pathology. One hospital administrator told me once when you start a hospital you need a kitchen and a pathology lab. The rest comes from there. Now this practice, which is really central to disease diagnosis for cancer treatment is at a crisis point. And the reason for that is that there are less and less pathologists, people are retiring from the field and not enough people are entering it. And on the other hand, you have a growing incidence of cancer worldwide. So there is a gap year between the capacity of the pathologists, the pathology labs to deal with the cancer coming in and it’s becoming worse because there is more and more cancer. We believe that technology like ours, the AI can come in here and bridge this gap.

Bill Pfeifer:

That sounds pretty fantastic. Okay, so radiology tells you something’s there and pathology tells you what it is and your system can help accelerate that process and improve the accuracy. That sounds pretty amazing. Now your latest announcement is for the second read system to diagnose breast cancer, but you’ve already established the second read name with your initial product release, which is focused on diagnosing prostate cancer. I believe this is the first AI system actively running in a live clinical pathology lab. Can you tell us something about where it’s running and how you managed to get it into a live medical lab? That’s a pretty amazing accomplishment.

Joseph Mossel:

So we work very closely with the second largest healthcare provider in the world. They’re called Maccabi Healthcare Services and they have the largest pathology lab in Israel and we work hand in hand with them in developing our algorithms and we use them as a test bed for our technology. But when I say test bed, this is not some kind of synthetic testing. This is actual running in a live clinical setting. And we made a bit of history here. As you mentioned, we’ve been the first company to deploy an AI based system in a pathology lab in a real live clinical setting.

So this is used on a daily basis to screen through pathology cases by real patients. So this was already a piece of history. What was amazing for us that for both products that we deployed very soon after deployment, there were already cases where we were able to detect cancer, which was misdiagnosed and when I say here misdiagnosed, I mean the case came out of the lab diagnosis benign, Ibex’s second read was able to flag that there was a high probability of cancer there and this was confirmed finally by a pathologist and they were able to revise the diagnosis. I think it’s pretty obvious to anyone listening that if you miss a cancer than this person is not going to get the right treatment. And I’m not a person who tends to over-dramatize here, but this can really be a death sentence in some cases.

Bill Pfeifer:

Certainly not the news that you want to get, but for sure you need accuracy and you need to get the bad news if you’ve got that coming so that you can do something about it. And speaking as a healthcare consumer, I would definitely appreciate anything that gives me better accuracy and tests that affect my life. Now, your site covers a lot of the technology of healthcare, computational pathology, lots of AI, of course, telepathology, but that’s the operational side of things. What does that mean to an average healthcare consumer like me?

Joseph Mossel:

I think if you’re a healthcare consumer and you have suspicion of cancer and it’s sent to be diagnosed in the lab, I think there’s a few expectations that you have. You want to get the right diagnosis, so that’s what we discussed before. Accuracy. You want to get it in a timely manner both because it’s important to start the treatment quickly but not less important that you’re under great emotional stress while you’re waiting for your results and you want to have them quickly. What you also want is given the accurate and timely diagnosis, you want to be put in the correct treatment pathway and Ibex can also provide insights here.

Bill Pfeifer:

That’s pretty amazing stuff. Now, you mentioned earlier that cancer incidents are expected to double by 2030 and the number of pathologists available will be decreasing as folks retire and not as many people enter the field. So you’re using AI to help increase the productivity and accuracy of pathologists to help close that gap. But can you talk a bit more about that actual workflow? What work does the AI pickup? What work does the pathologist still do? I assume we still need pathologists, but hopefully this will help close that gap some.

Joseph Mossel:

Yeah, so the second read product there is a safety net for the lab. It’s after the fact, after a pathologist has already reached a diagnosis. It’s a safety net. You feed in the diagnosis, you feed in pathology slides of the case into the algorithm, and if the algorithm thinks there’s a high probability of a diagnostic error, it will raise a flag. The first read product, which is here to drive efficiency comes before the pathologist. That’s why it’s first and what it does it precooks the case for the pathologists. It marks which slides are benign, which are cancerous. It does degrading in advance. It prepares a draft of the pathology report so it helps the pathologist do their work much, much faster and more accurately and that’s where you see the dramatic efficiency gains that technology like ours can bring to the market.

Bill Pfeifer:

So can you talk a little bit more about how it works first read versus second read?

Joseph Mossel:

The second read really comes to deal with the question of diagnostic accuracy. It doesn’t improve the efficiency, the productivity of a lab. First layer it’s very important because first and foremost patients want to get the right diagnosis. But if we look at our products which are in the development, first reads products, that’s where the real efficiency gains come from.

Bill Pfeifer:

So first read is doing a run through and highlighting potential trouble areas so the pathologist can focus just on that space and then does some of the paperwork on the back end so that they can speed up their work. Where, the second read is double checking the diagnosis that’s already been provided.

Joseph Mossel:

That’s very well put.

Bill Pfeifer:

Thank you. Now as we were preparing for this episode you mentioned the evolution from traditional to digital pathology. Doing high resolution scanning of glass pathology slides into very large files and how that’s creating challenges for IT to keep up with the compute, storage and network requirements. Is that an overall industry change or is the transformation to digital pathology driven primarily to support AI? So specifically are we digitizing the industry and that’s making the AI possible or is the AI driving the digitization of the industry?

Joseph Mossel:

So I think the two are going hand in hand and I’ll talk just for a moment about these IT challenges. Ten years ago what we’re doing today would have been really science fiction because of these IT challenges. Digital pathology is happening independently of AI. It allows for a smoother workflow. It allows for digital archiving. It allows for doing pathology remotely. So there’s a lot of benefits to that. On the other hand, as you alluded to, it does create IT challenges. The file sizes are huge here, so there are significant storage and networking requirements. So now you have this part of the hospital which was really working with microscopes and glass slides suddenly moving to the forefront of IT. So, that’s part of it. But what I sense and what I see from talking with people in the market is that the killer app for digital pathology is AI.

Digital pathology in a way enables now to do pathology in a bit of a different way, but the AI really changed the practice profoundly. It changes what it means to be a practicing pathologist. It changes the way a pathologist will be trained. It changes the best practices of the discipline. I think both us and the practice, we’re still in early stages of understanding what will it mean, but I have to say pathologists which see our system in action, they heard about it, but once they see it, that blows them away. They realize their practice, the way they operate is going to change completely.

Now a question I get asked often, especially by pathologists, maybe a bit worried is whether AI with the products like ours is going to replace them and I think that might be true someday, but I don’t think that day is coming tomorrow. What I am very much convinced that pathologist who will not be using AI will not be practicing within the next decade. Much in the same way as autopilots and planes hasn’t replaced the pilots, but no one will be wanting to fly in a plane today, which is not using an autopilot.

Bill Pfeifer:

It’s amazing to think that Ibex is changing the entire industry of pathology. Some of it’s probably changing on its own, but I love talking to you knowing that you’re on the forefront of driving this entire change. So now you’ve got prostate cancer and breast cancer diagnostic capabilities with the second read system and you’re moving toward that initial scan to help improve the efficiency of pathologists with the first read scan. What’s next for Ibex after that? Once you’ve disrupted this piece of the industry as though that’s not enough, what comes next?

Joseph Mossel:

The natural evolution for Ibex is to go and do further indications, more tissue types. So we want to be really an end to end solution for the labs. So supporting at least the most common cancer types and the engineering team, the data science team back in Israel are working very hard to churn out more and more algorithms and we’ve got that pipeline pretty streamlined by now. So, that’s part of it. Another element is asking a different question than what we’ve been discussing so far. What we’ve been talking about so far is looking at the practice of pathology and asking the question, how do we make this more accurate, higher quality, and how do we make it more efficient?

But there’s another question one could ask is how do I use the algorithms to glean new insight from the data? How do I combine the image analysis, the AI together with other data, which is available to introduce predictive algorithms, algorithms which can help determine prognosis and can help guide treatment selection? So where we see ourselves eventually as being an end to end solution for the cancer pathway, more accurate diagnosis and a more timely manner with the best insights to select the right treatment.

Bill Pfeifer:

This is a huge challenge you’re taking on. This is pretty amazing and I love the success that you’ve already seen. I think I speak for all of our listeners who wish you great success, Joseph, if I ever have to go through scans looking for cancer, I certainly hope I have the fastest and most accurate results possible in anything you can do to help improve my odds would be really appreciated. Thanks so much for spending the time with us and giving us a view into your world. We definitely appreciate your insights and look forward to hearing more about what will be coming out of Ibex’s AI assisted cancer diagnostic program over the next few quarters and years.

For those of you who enjoyed this podcast and want to know more about what’s happening with Ibex you can find out more on their website at ibex-ai.com. That’s ibex-ai.com. They have some great overviews of what they’ve been doing along with links to other related resources and you can find more information about The Next Horizon at www.delltechnologies.com/nexthorizon, including future podcasts and some great technology content that we’ve shared from the smart folks over at our office of the CTO.

Thank you all for listening to The Next Horizon, a Dell Technologies podcast. We appreciate your time, interest, and attention. I hope you’re as excited as we are about the great innovations that are coming out of cutting edge companies like Ibex. Be sure to subscribe to the podcast, either through your favorite podcast app or through the website at www.delltechnologies.com/nexthorizon so you don’t miss any great new content and I look forward to seeing you again for upcoming episodes.

I’m Bill Pfeifer, and this is The Next Horizon.