Smartphone, smartwatch, smart TV…internet of things (IoT) and artificial intelligence of things (AIoT) is ubiquitous. But did it already make it into any of the digital pathology devices?
Oh yes! There is a smart whole slide scanner out there.
In this episode, I am talking with Don Van Dyke, the chief business officer of Bionovation Biotech. Bionovation holds a patent to a potentially revolutionary scanning technology powered by AI. Due to the ability of the scanner to predict the 3D architecture of the scanned tissue in the Z-axis the device is able to dynamically adjust camera focus exactly to the surface of the specimen and scan it ca. 70x faster than the classical whole slide scanners.
Not only does it make the scanning faster, but eliminates the necessity of Z-stacking when scanning smears and cytology specimen. What is more, it is now possible to obtain high magnification images (80x and 100x) fast too, which practically removes most of the digital pathology hurdles for hematopathology and cytopathology.
To learn more about Bionovation offer visit: http://www.bionovationimc.com/
Transcript
Aleksandra Zuraw: [00:01:35] Hi everyone, and welcome to the podcast. Today. My guest is Don van Dyke, the business development officer of Bionovation Biotech and Bionovation’s goal is to use innovative technologies to improve digital imaging in pathology.
[00:01:51] Hi Don, how are you today?
Don Van Dyke: [00:01:53] I’m doing great. It’s great to see you.
Aleksandra: [00:01:56] Good to see you.
[00:01:57] And thank you so much for joining me on the podcast. And let’s just start with a brief introduction that the listeners about yourself and your company. Bionovation.
Don: [00:02:08] Yeah. I began my career a long time ago selling laboratory products into pathology labs and all clinical laboratories 70,000 products that could turn an empty building into a clinical laboratory.
[00:02:21]Did that for eight years with American hospital supply and then joined a startup company in quantitative digital image analysis called Bioimage. And we were acquired by a couple of different [00:02:35] companies. We became dominant in proteomics and certain areas of genomics image analysis.
[00:02:40]While we were owned by Millipore, I put together a partnership and bought Bioimage from Millipore and ran that as CEO, I was with them for 11 years, last three years as CEO, eventually that company merged with another company and went public on the NASDAQ. Then I spent five and a half years in artificial intelligence with Cloud Pharmaceuticals.
[00:03:03] Where we did target-based drug design using AI from Duke university local here to, where I am in Raleigh, North Carolina. And then I discover just about a year ago, came across a Bionovation which combines pathology, digital image analysis, and artificial intelligence. I’m the business office, the company was founded.
[00:03:25]And founder and primary owner is Yin Yuefeng who is located in San Diego. And he had the founding of the company is really [00:03:35] interesting, but so he is, a technical person, an inventor, very excellent engineer. Has put up a laboratory instrument successfully in the past. I’ve always been on the business side of scientific instrumentation.
[00:03:47] So we fit together that way.
[00:03:48]So you say the company’s main location is in San Diego?
[00:03:52]Yes. The company, Bionovation Biotech incorporated. It’s founded in 2017 in San Diego, it’s registered in California. Yin of course is from China, from Suzhou, and has business interests in Suzhou and also that’s where most of the R&D and production and all that sort of thing takes place for Bionovation, but it’s an American company.
Aleksandra: [00:04:18] So what does Bionovation offer? What are your products or services or both?
Don: [00:04:25] We have, novel innovative whole slide imager for pathology slides. And the reason why it’s innovative has to [00:04:35] do with the core technology that’s been patented by Yin. It allows extremely high-speed high magnification scans to take place in a short period of time.
Aleksandra: [00:04:44] So by imager is a scanner, right? It’s a whole slide scanner or other types of specimen as well.
Don: [00:04:52] The whole slide they’re large slide scanners, but a scanner, but the Bionovation CSFA 800 is designed only for pathology use.
Aleksandra: [00:05:02] Okay. And you say it’s high speed. How high is the speed? How much faster is it than the other products, the other whole slide scanners that are on the market?
[00:05:12] Don: [00:05:12] It’s an interesting story. Up until Bionovation, there were only we’re aware of, there were two patents having to do with the automated focusing and scanning of pathology slides. The first one we’re aware of is from Hamamatsu. Some time ago. And then the next one from Aperio, which is Leica, and all the scanning, whole slide scanning we’re aware of has to either [00:05:35] use those or it has to license those patents or use those patents.
[00:05:38] Bionovation Yin invented a new system that was different enough to become patentable. And it’s so novel. The core technology allows the system to predict to actually find what the curved surface is the true surface of the microscope slide. And by mapping that out the scanner can read very fast because the Z-axis knows where the next step ought to be.
[00:06:04]And so the scanning takes place up to 70 times faster than the current state of the art.
Aleksandra: [00:06:11] So what’s the time for 1.5 centimeters x 1.5 centimeters? This is like the standard measurement, of whole slide imaging. How fast does it scan?
Don: [00:06:25] Yeah. 20 seconds, something like that.
Aleksandra: [00:06:28] Okay. That’s fast
Don: [00:06:29] Well, let’s go in a different direction. Let’s take it out to the end. Can do a whole slide [00:06:35] image of a peripheral blood smear at 100 X oil in three minutes or less
Aleksandra: [00:06:42] 100 X in three minutes. Okay. That is fast. Indeed.
Don: [00:06:46] Yeah, it’s the whole slide.
Aleksandra: [00:06:48] First of all, I’m not aware of any other scanner that does 100 X and there are some,
Don: [00:06:57] yeah, they take a long time from what I’m told. So that opens up a whole new way of looking at whole slide imaging. And we’re exploiting and teaching about what that can mean to a pathologist about the accessibility of high magnification high-speed imaging.
Aleksandra: [00:07:14] You’ll have also the 20 X and 40 X magnification in the scanner as well.
Don: [00:07:19] Yeah it’s a, it’s quite a compact system. if it is on a benchtop, but it has a five-place, most pieces 20, 40, 60. 80 a hundred oil, you can, their standard objectives. So pinch those around, but [00:07:35] the usual configuration is 20, 40, 60, 80, and one hundred.?
Aleksandra: [00:07:38] Okay. And so what is your mission with this product? What is, where are you going with this, except that it’s faster?
Don: [00:07:48] It’s a high speed into Into high magnification is really our mission, but there’s a secondary mission that provides data that provides image data, but we’re also committed to providing answers immediately on the fly.
[00:08:05] So there’s a bit of a breakthrough with this very high magnification high scanning capability, but we also incorporate AI hardware right into the scanning platform. And we actually do deep learning calculations on the fly while the scan is taking place and deliver answers immediately.
[00:08:25] High-speed high magnification and immediate results.
Aleksandra: [00:08:29] So you have the image analysis of the slides that you’re scanning [00:08:35] inside of the microscope, and this is your proprietor, you, which analysis, or do you have an open API to be able to use other software from other vendors?
Don: [00:08:44] And this is where Yin and I really agreed without ever knowing each other. I came from AI for five and a half years. And It’s my opinion and the opinion of others I work with that AI will become ubiquitous. Okay. It’s going to become like an Excel spreadsheet. And so encouraging AI to be ubiquitous is the right direction strategically to go, in Yin and my opinion, therefore, the image files are open architecture. Frankly, proprietary architectures for images are stupid.
Aleksandra: [00:09:19] Which is the case in pathology. Everybody has their own proprietary format unilike
Don: [00:09:25] radiology
[00:09:26] yeah, it’s simply greedy. And there’s no reason why a scanner can’t return any image file format that you ask it for.
[00:09:34]11 years of, [00:09:35] Bioimage, you can put out any format that you want. Let’s put out a format that the users would like to have, so we let them choose which image file format or open architecture. Okay.
Aleksandra: [00:09:45] And this is a capability of the scanner? You can choose the file output you want?
Don: [00:09:50] you can choose TIFF, you can choose .svs . Yeah, it really is trivial. It should be trivial for everybody.
Aleksandra: [00:09:56] Okay. And seems like such a big deal. Everybody is making the right own file format. And that’s very confusing, especially for downstream analysis for computer scientists. Not anymore for the users because you can download any viewer and you have viewers to support many file formats.
[00:10:17] But what I have heard from computer scientists that are doing AI, that they’re designing algorithms to analyze, they have to troubleshoot first, if they have images from different sources to make a common denominator for those
Don: [00:10:32] well, okay. Let’s look at the electronic [00:10:35] health records track records.
[00:10:37] Okay. And it’s a similar mindset where the early deliverer of electronic health records wants to keep the customer locked in. And not allow interoperability with other systems in order to increase the short-term revenue and the profitability of the firm. That actually is very noxious towards the ecosystem of electronic health records where we see in other disciplines where interoperability of data types is accepted.
[00:11:05]And in EHR it’s moving heaven and earth to get interoperability and it’s preventing innovation from taking place. And it’s the same mindset that goes with scanners that want to lock you into the file formats. We’re a little tiny company but at least we’re doing our part to make the images open and available and the same thing with the AI architecture.
[00:11:25]So we believe that AI or deep learning, I should say will become ubiquitous, particularly in image analysis and management. And therefore [00:11:35] we certainly provide an onboard solution, but if you like part of what we have and you want to add to it if you’d like none of what we have, do you want to change?
[00:11:46] Do you want to do your own. That’s fine. We’ll enable that group or that organization to do the training sets to run them on the system and to have their own particular version of deep learning that they want to use. Not what some company has told them that they want to use.
Aleksandra: [00:12:03] I’m asking because many people already have image analysis solutions that are working for them, some deep learning-based.
[00:12:11] And I was wondering that would be possible to integrate with your system and from what you’re saying it is possible. So two quick questions, how big are you, and how many slides fit in the scanner at the same time?
Don: [00:12:23] So all told worldwide we’re somewhere between 45 and 50 employees. We are kinda lean on the business side and the executive side you’re here in the States, but that’s the size of the [00:12:35] company.
[00:12:35]So the current systems that we have out right now are two slide cassettes. So you could swap out two slide cassettes in that format. By the end of December 2020. We will have a 30 slide cassette system auto-loading system. It’s being, it’s actually up for FDA approval in China CFDA right now.
[00:12:58]And then in, in 2021 we expect to go to in the thousand to 1200 slide loader
Aleksandra: [00:13:04] Oh, wow. That’s a lot. So who are your customers at the moment?
[00:13:09] Who are your current customers and also your target customers?
Don: [00:13:13] Right now we’re in an evaluation in a clinical Humana pathology section at Newsweek ranks as one of the top five hospitals in the world. And that evaluation has been going well. They’re seeing that we can do things that haven’t been done before and there’s the gate.
[00:13:31] There’s a gate right now. If an [00:13:35] organization in the pathology group or hospital system wants to go completely digital. There’s a gate because cytology and hematology, pathology or hematopathology require higher magnifications that are generally available in the standard scanners. So what we’re seeing is the high mag and the speed can fix that problem for.
[00:13:59] For organizations that want to go fully digital. And so that’s where we’re focusing first.
Aleksandra: [00:14:05] That was exactly what I wanted to ask next because you have the technology that predicts the surface. It’s less data-consuming to do the Z stack. You’re just going on the surface of the side. You don’t have to do plenty of these stacks at different levels.
[00:14:26] You probably can, but you don’t have to have a high-quality image from one scan.
Don: [00:14:33] Exactly. Let’s do the [00:14:35] absolute best job we can on the first scan to get and where the peripheral blood smear, it’s not going to be thick. Okay. So do that as well as you possibly can, but you want to do the Z stacks. Of course, we can do Z stacks and we all, and also have some. nimbleness about how to handle images. There are ways to do Z stacks and reduce the image size and retain the information and the detail of the image. There are ways to, at very high magnification and still keep manageable file size and retain all the information and detail on the images and so forth.
[00:15:06] So it, it’s exciting what you can do if you apply all the tools that are available.
Aleksandra: [00:15:12] So where is your ideal place to be? And let’s say in five years you have the company fully developed, fully functioning. Where would you want to see your scanners everywhere where scanners are there at the moment or you have a little bit more niche approach to that,
Don: [00:15:31] While the platform has a [00:15:35] really compelling envelope. It’s a benchtop. It can be self-installed. It doesn’t require our technicians to come and self-install, but it has precision down to 0.2 microns and X Y platform. So any of your 20, 40, 60, 80, you can do all those magnifications yeah. It could be used widely. There are certain disciplines though, in pathology where the current market leaders have earned their market leadership position by doing things about enabling workflows, about working well in a production environment where many pathology, clinical pathology laboratories really are a production environment, particularly in the for-profit labs.
[00:16:15]And they do a really good job of being the Amazon, crank out the product as quickly as you can. And as low-cost as you can. The opportunity for us, we seem not to be, to go directly against that, particularly at the size and development level that we are.
[00:16:31] So we’re going to the places where people need us. We do [00:16:35] things that people cannot do so far that’s our beachhead. Okay.
Aleksandra: [00:16:39] And probably not for the hardware, but for the software inside. You must at some point, work with pathologists or have a vision, how to work with them. Maybe you can tell me about it.
Don: [00:16:53] Great. So we have several AI I should say deep learning computer vision models that we’ve already developed. Those have all been developed in Asia. With the co cooperative deep cooperation of pathologists and practicing laboratories in Asia. Right from the beginning the development of our peripheral blood analysis, the bone marrow, the cytology, and now we’re working on certain cancers.
[00:17:15] That’s the R &D is primarily taking place in Asia with local pathologists there. Since I’ve been here, we’ve begun to make relationships with some leading universities and cancer centers in the United States and in Canada where we’re partnering with the, with some thought leaders to [00:17:35] have them evaluate and publish and give their inputs to us about, what the particular strengths are of the platform.
Aleksandra: [00:17:43] And what were the problems that you encountered while developing your offering while making this device something that you had to overcome that you did not expect? You’ve been in the industry for a long time plenty of hurdles, but was there anything that was unexpected? Does she tend to pay for it and do something different?
Don: [00:18:08] Personally, for me as a manager, I’ve been away from pathology for some time. So I had to catch up to what the current environment is. , that took some time. It turns out that the paradigms that are driving pathology and digital pathology and remote pathology.
[00:18:26] Outside of most of North America or even stay outside of the United States, those drivers seem to be somewhat different significantly different. [00:18:35] And so assumptions that we make about the marketplace outside of North America and outside of the United States don’t actually play as easily within these more highly regulated environments.
Aleksandra: [00:18:47] So what are the differences? Is it the regulation or other directions of where the needs of pathologists are?
Don: [00:18:56] It has, so the two things, one is regulation is such that, HIPAA is a relatively new thing still in the United States. Healthcare just has this really clunky way that they tend to embrace new things.
[00:19:11] Okay. So HIPAA suddenly becomes the set of handcuffs that, that, all of a sudden, and we have to get approvals from various levels of it. And some of the logic makes sense and some of the logic doesn’t make sense and so on. So yeah, you have that portion, everybody has to deal with it.
[00:19:28] And that’s the way that it is. The second thing is that I think I, I think regulation, I’m [00:19:35] not sure I’m still coming back, but I think regulation or something has made the United States and North American pathologists more or less eager to embrace digital pathology. And so the combination of those two things makes this marketplace different.
[00:19:52]Even in Canada Canada digital pathology is Hey, let’s go.
Aleksandra: [00:19:56] I worked in Canada in Montreal for over a year and yeah, they had a fully digital hospital. I met them at two US conferences and you’re talking about things that are not happening in the US unless you really. Are very invested in that.
Don: [00:20:16] My early placements to my early evaluation placements are in Ontario, frankly, because
Aleksandra: [00:20:23] networks and I guess because they have plenty of remote places. So does Norway, where they also adopted telepathology as the first country [00:20:35] in the eighties, early nineties. And. But yes, you’re right, totally different mindset and things are allowed that here, people are still very cautious about.
Don: [00:20:46] And I think also, and here’s me image guy, scientific products guy. Okay. I think the industry ourselves or industry as that existed is also, it thinks in a direction. Like this: let’s try to take what the pathologist is doing. Put it on a platform and they’re not accustomed to. Okay. And then try hard to replicate what the microscopy experiences.
[00:21:15] Okay. So try to replicate what they already can do on a microscopy platform. How about. We do things for the pathologist that they can’t do. Okay. That hasn’t been done and create the pull through. Okay. Wait a [00:21:35] second. You’re telling, I ha honest to goodness yesterday in a conversation I had, I met a pathologist.
[00:21:41] Are you kidding? You process 800,000 white cells in a bone marrow smear while I’m watching it. I’d say, yeah, we do 5,000 cells a second that, encourages a pathologist to say, wait, I want to go do something new. Okay. I’m not trying to replicate my microscope experience. I want to do something I can’t do
Aleksandra: [00:22:04] This probably refers to the AI algorithms that you have in the scanner. What do you have there? What solutions do you already have that are ready to be purchased? I assume they come with a scanner or you can get them with the scanner?
Don: [00:22:16] At the moment we’re being generous. Cause we’re packaging them all on a single price.
[00:22:21]Anyway it’s very attractive and no maintenance agreement. There’s no software for the first year and just, it’s just here’s your system. But anyway, we have a particular emphasis on hematology and of pathology. And there’s a good reason for [00:22:35] that. If you don’t mind me expanding, talk about our founder, Yin He was a technology lead in a company. Called Xitogen, it’s spelled with an X I T O G E N. So Xitogen from Suzhou produce flow cytometers and so Yin spent several years developing flow cytometers for Xitogen. And they, Xitogen was eventually acquired by Beckman Coulter for pretty good money. Like a few years ago. But that’s where you develop this whole concept of, flow cytometry which is taking a single image, a single cell images at very low resolution on different wavelengths. What could happen if you had a full resolution on those cells, and did image cytometry on these images.
[00:23:24] And that’s why the CSFA 800 is called the image cytometer and that’s, what’s been driving this. So the AI-first was developed for hematopathology, [00:23:35] obviously, because that’s the direction that we began with. Cytology is included in various flavors. And then, so we’re working on different solid tissue applications.
[00:23:46]I have a partnership going right now with an Ivy League university where they’re. Investigating virtual IHC on tissues. There are a lot of places to go with tissues. It’s not as clear as cytology or hematology. We would like to recruit pathologists and groups and universities to work on this with us in an open architecture to come up with community-based or community-driven solutions that would work great for us.
Aleksandra: [00:24:11] This is fascinating. And is there anything else that they forgot to ask you and you would like to tell the listeners?
Don: [00:24:18] I’ve said a lot. I think that the key issues are number one, the core technology is only limited by the data transfer rate.
[00:24:27] So we’re now up to a nine-megapixel camera that can do 290 frames [00:24:35] per second to bring that scanning speed even further down. So we think the technology is going to be durable for a long time. As that goes we think that making this accessible to people in an inexpensive way , in an open architecture way.
[00:24:48]In a way that’s easy to set up and to maintain is the wave of the future. I guess I’ve said all that. So I dunno if I can add any more.
Aleksandra: [00:24:55] Thank you so much. Thank you for joining me for this interview for letting the listeners know about this innovative technology, where can we find you online?
Don: [00:25:03] You can find us at Bionovation.
[00:25:06]Just Google B I O N O V A T I O N. We’re there and we’re all over LinkedIn.
Aleksandra: [00:25:13] Okay. I’m going to put the website in the show notes. Thank you so much for joining me and have a great day.
Don: [00:25:20] It’s a real privilege. Thank you.