5: HistoWiz: fast histology and an image database for mining from Brooklyn NY w/ Ke Cheng

Hi, Key. It's so nice to have you on the podcast. Welcome. Thank you. Hi, Alexandra. Thanks for interviewing me.

Let's start with telling the guests about yourself and about your background a little bit. Sure. So I'm a cancer researcher with over 10 years of experience on mouse modeling of human cancer. So I worked at Pirabola Pandolfi's lab at Harvard during my PhD. And I worked with Doug Hanahan, who is a very famous cancer researcher in metastasis and endogenesis for my

postdoc at Swiss Cancer Center. So that's where I started the company. So I was mainly in academia. You are the CEO of HistoWiz. What was your motivation to start the company? So my main motivation came from my own frustration of cutting tissue. So I was cutting slides.

myself at the core facility and I realized that there's a huge bottleneck for histopathology for researchers and the turnaround time usually is over two weeks and as most of researchers know histopathology is the last step of any experiment before publication. It's very frustrating to wait for results so long. So my colleagues and I were always looking for an

sourcing company to do it with good quality and it's very hard to find that, especially in Europe. So I started Histowiz to help researchers accelerate histopathology and find a cure for diseases faster. Tell us about the company. Where are you located and what are your products and just a picture of what do you do at Histowiz?

Aleksandra Zuraw (02:22.722)
Yeah, sure. We are located in Brooklyn, New York. A lot of people ask why we do it Brooklyn. there are actually Boston obviously has a lot more biotech companies. But in New York, we have a lot of great academic institutions such as MSKCC, NYU, Cornell, Columbia, Mount Sinai, Albert Einstein. They're all there. So we started from academia.

So we are located in a biotech incubator. It's one of the few wet lab space in New York City. And we have customers all over the world, but we started from New York. Okay. And how many people are you now at the moment in the team, more or less? We have around 20 people.

Because we automate histology for researchers, most of the work is done by machines. So we have a lot of equipment in the lab and not that many people as you might have guessed. Okay. And what is the mission of Histowiz? So that's a great question because it's definitely a mission-driven company. When I started the company,

You know, I basically left academia to start this company. I've never done any companies before, but I really believe the value of the company is building the centralized pathology image database. So not only is histology very frustrating for researchers, there's not a single centralized pathology image database. When I was doing my research, there's nothing

that you can compare histology data similar to genomics data. So there are genomics databases, there's EHR databases, but pathology database is the missing link, I think, for disease research and for finding cures. And so I think this is really important and valuable to build. So digitization is key for our company.

Aleksandra Zuraw (04:49.486)
We, as a company, we know the biggest pain point is tissue processing, but we actually, since the first day, we decided that digital pathology is very important. And back in 2013, when nobody was really scanning slide, we started.

every single slide that our lab produces. And every single slide goes into a centralized database. And we have clients who want to contribute to the scientific community. So they give us the rights to their data and images after one year for an upfront discount on their histology service. So most of the academic labs want to contribute to the community. And they give us the option to

have their data available for this centralized database, for other people to do data mining. Okay, so they do have a good incentive to do that, to contribute. That is great. So two reasons you said, first, why you started, first, it takes too long to do HISTOS, second, there is no centralized image database and both of these

Fast Histology and Central Image Database HistoWiz is providing? Yes. So what do you wish you had known before you started? That you, of course, know now, but it would be easier if you had known it before. I think in the beginning of the company, we were trying to do too much. the long-term goal is always to use this database to...

generate value for the scientific community. But in order to get there, we really have to focus on the histology service to make it scalable. And I think in the beginning, we were trying to optimize our histology workflow. We were trying to develop the database. We were trying to develop AI tools. But I think I wish I had

Aleksandra Zuraw (07:10.146)
been more focused on the histology service to make it unbeatable and make it as fast and convenient as convenient for researchers because that's really the biggest pain point. As scalable as possible. Right now that's really our focus. I think in order to get to database and AI tools, we really need to make sure that the service is running like a wall.

machine. I think this is also a kind of the entrepreneurial process where you try out a couple of things or wish you can do more and then you focus on the most important like you did. Any failure that you had that set you up for success now? It ties a little bit in what you already said that spreading yourself to thin at the beginning then to focus so that could be it but any like

particular event? To answer this question, I can say that I made so many mistakes that every single one contributed to making a better company. So there's not a big mistake, but there are so many small mistakes we have made. But that's the part of the improving the company. And when did you start the company?

So I started operation in January 2013. The company was incorporated end of 2012. We started operation in Brooklyn beginning of 2013. And the first service is just histology service and the clients will see the slides online. So 2013, you started and you're still in business. So all those mistakes.

did contribute to your success, whatever these were. We're still around. Exactly. So maybe you can tell the listeners a little bit about your service. What do you offer? Yeah, I mean, the service is definitely valuable. Otherwise, we wouldn't be in business anymore. And I think, being a researcher myself, I understand the frustration and I understand our customer base really well.

Aleksandra Zuraw (09:35.49)
So what we provide is the fastest turnaround time histology service for researchers, whether you're doing HNE or special stains, IHC, IF. You can malice the tissue in formalin, and in as few as two days, we'll get the slides, examples, processed, embed, caught, stained, and scanned.

And you can see the results after two days on any computer or mobile device. High quality, 400x magnification images of your whole slide images. So instead of looking under a microscope, you can see these beautiful digital slides on any computer or mobile device. And this really allows collaboration. So clients come to us not only for the fast turnaround time, but also for this

amazing platform, web-based platform, to access their results and share with their collaborators, get a diagnosis from a pathologist. So we started a new service that's telepathology, basically getting the slides interpreted by a board-certified pathologist in 2018. And this is

a new service that's really taking off, we realized like all our clients that need the tissue process also want to get their slides interpreted. So that's a new service, pathology service that we offer. wow. So basically, let's say a company doesn't have a pathology lab, but does pathology experiments not only can they do the wet lab service,

at your place, but you also have this pathology service available. Yeah, yeah, exactly. this is great. I actually have not heard of any other company providing this in that form. So, you know, we have a network of over 60 board certified pathologists with a wide variety of subspecialties that can help our clients to do analysis and make diagnosis.

Aleksandra Zuraw (11:56.174)
It's really one of the unique advantages of our platform because all the slides are available online. We can provide consultations digitally. So who are your customers then? So our customer we have already over 2,000 clients from all over the world from over 500 labs.

over 200 institutions and over 100 biotech environments, such as SKCC, Harvard, Stanford, MD Anderson, Jackson Lab. So in terms of companies, have General, Shire, Jensen, and a lot of biotechs such as CRISPR, Editas.

and a lot of new clients have started recently. So what are the needs that can be met by your company? What would your target customer be? So let's assume I could be your customer. What am I looking for? Yeah, so the value we bring is the speed and quality. Because we're fully automated, the results is

are very reproducible. It's the scalability that you can send hundreds of samples and expect them to be done in three days. But I think also one of the unique things is that we have expertise in-house, from our IHC team and also from this network of pathologists. They're remote, but they can provide

consultations and help study design and generate reports for IND submission, clinical studies, this is really hard to find. Okay, okay. So no need to have an IHC lab or histolab, nor even pathology team. It can be all done as histories. Right. That's why we have so many biotechs that started recently beside these big pharmas.

Aleksandra Zuraw (14:19.702)
As I mentioned, a lot of biotechs in Boston, South San Francisco, Corvius, Nectar, Fibrogen. I can very much see that because they do not have pathology groups, but they do need the pathology expertise and work for the substance development or drug development, whatever they're working on. So what differentiates your company from the competition?

Great question. We are focused in preclinical and research studies. That's my personal background and we don't process any human samples for diagnosis. We process animal samples. So from zebrafish to patient derived xenograft and organoids, we have a lot of experience with different cancer models. And I think this is

A lot of biotechs and academic clients really need these type of services, focus in pre-clinical samples. For biotechs, I know that the discovery phase where lead compounds are tested, that's the first stage of drug development and a lot of money are spent on this discovery phase. So if we can help them accelerate

this part, we basically help them save a lot of money down the road and help them make better decisions. Pharmacists and biotechs also want tox studies like GLP studies, and we can provide that. many academic groups want to do also translational work, which is going from animal models to human samples so we can

process human samples as long as it's for research, not primary diagnosis. So why digital pathology? I think when I started the company, even though it wasn't as hot as now, I really saw the need for researchers to collaborate using digital pathology. As you know, FDA has finally approved the use of digital pathology for primary patient diagnosis.

Aleksandra Zuraw (16:47.882)
up until now or up until a year ago when this was approved, there was no money in it. So nobody else was really digitizing the slides except us. I really believed in it because I know researchers have their own funding and they will pay for it just to make to share their results to publish these digital slides so that the reviewers and readers of their manuscript can even see unbiased results and

having higher reproducible results is really important for the scientific community. What do you think is the most valuable contribution of historians to the field of digital pathology? I think the dream that I have and that's what that's why I started this company for is really the centralized digital

centralized pathology image database we're building, even though we haven't really commercialized this product, this is a new product that we're planning to commercialize now. It's really over five years, almost six years of work that we have accumulated.

We have over a hundred thousand digital slides in our database from clients all over the world, from all kinds of species, organs, biomarkers, experimental treatment conditions, genotype, and nothing like this exists. So you could really do data mining and discovery in using this database instead of having to repeat someone else's work and know that

one researcher might be looking at a specific phenotype like metastasis and they would not look at other phenotype, but it could be very interesting for someone else to look at it. And how do you work with pathologists for developing your services? So pathologists are our main pillar of histo-wiz because they bring

Aleksandra Zuraw (19:08.748)
They're from different fields and different subspecialties for our clients. They are consultants for InstaWiz and they can view client slides online and give a diagnosis. So what we provide on our platform is a matching system. So we have a process to distribute these work

and also match the client's request with a pathologist with some specialty. Do you also have pathologists on your team? We do have a chief pathologist, Chris Gibson, who has a lot of experience with pharma and GOP study. He worked at Merck before and developed the Kichuda.

antibody and over 10 years experience in IHC and biomarker validation. Wow, so they are part of your team so you have the expertise there on site to create products and services. We have some amazing pathologists from Harvard, NCI, Mayo Clinic, Johns Hopkins, even the FDA. They have all kinds of subspecialties from

impasse to tox pass, breast, neural, anything you can think of. I want to ask you about the innovation. So you are solving the greatest pain point, like you said, of all research institutions, be it academic or commercial, fast histology. What is innovation in your company? So we have developed

A streamlined process was fully automated equipment to process tissue in as few as two days. We have a proprietary QC process to check for sides to make sure they have no wrinkles, folds, air bubbles, all within this turnaround time. this is crucial for digital pathology. Not everybody is aware of that, but

Aleksandra Zuraw (21:35.464)
Yes, it's crucial. The quality of the result is really high. We have a proprietary laboratory information system web-based to track the results at every stage. It's useful for auditing purpose for GLP. I think the biggest innovation is this database and the pathologist network. Our pathologist network is one of a kind and this is the first online pathologist network containing

comprising of veterinary pathologist, giving diagnosis remotely. you know, remote diagnosis is not FDA approved for primary patients, but we can do this for animals and for research studies, no problem. And do you incorporate AI, artificial intelligence, into any of your products or services? Yes, so we are working on

few AI tools, but most of these tools are not going to be, we're not focused on getting FDA approval, whereas we're developing AI tools mainly to improve our service workflow and improve our speed, our quality, such as an automated slide QC tool, such as a slide auto tagging tool. So one of the

very interesting tools that we have developed and published recently at a computer vision journal is the auto tagging tool. So this tool is solves a problem where each farmer have hundreds and thousands of untaxed slides and they would, it would take them a lot, a lot of time to

manually tag them based on stain type, organ, species, lesion, and cell type. So because we have the world's largest online preclinical pathology database with over 100,000 slides, and these slides are already tagged with stain types, species, and organ, we have a lot of training data. So we use these data to develop a tool to automatically tag slides based on these three criteria, and we have shown

Aleksandra Zuraw (23:59.854)
extremely high accuracy of tagging, especially for stain types such as H &E, we have 99 % accuracy for different species like monkey, mouse, human, pig, rat, zebrafish. They all have close to 95 % accuracy from a pilot study and also for different organ types such as bone, brain, breast, colon, heart, et cetera.

They all have over 95 % accuracy. So this is a very interesting tool that can help us to tag the slides in our database, maybe help farmers to also do this automatically instead of pathologists manually doing it. And using this algorithm that we've developed, it's called Patch Transformation Module.

we were able to find similar images as well. So our model predicted images much more similar than using the ResNet-50 model, which are used in other, for general images, used on Google. Like the dogs and cats and flowers and dogs and cats.

you have an algorithm to look for similar pathology images, you don't find similar images, but using our algorithm, it's very highly similar, especially in the past. The publication about the tagging tool, when did you publish this? So we just published this paper. Congratulations. Yeah, in August. this is published by...

The paper is called Patch Transformer for multi-tagging whole slide histopathology images. I will place the link in the show notes. Great, thank you. Let's do this one. The business world and the digital pathology world, let's be honest, there more men than women and definitely more male CEOs than female CEOs. Can you tell the listeners a little bit how it is to be a female CEO?

Aleksandra Zuraw (26:21.632)
of a digital pathology company in New York on top of that. Isn't this true for all the companies that there's more men's CEO in general? It's in general in business. That's how it is. I think, I mean, it's not surprising. Digital pathology is very innovative field. And I think being a female CEO,

I'm proud to be that. definitely think that empowers women to work in a field that's innovative. But this just shows that anything is possible as long as you work hard to follow your dreams and don't give up because it's really hard and it's

took us a long time to get here and build a good foundation, a good customer base. And I think it's really important that we have, our value is to serve our customers. We really believe that, you know, even as a leader, we have this servant leadership instead of military style leadership. We really think that as a CEO, I,

work every day to help our team to get their work done and get the work completed on time. And then in turn, our team pass on this value and these qualities to our client. And so our client is obviously the most important part of our business. And without them, nothing, none of the progress is possible. So

You know, I work to help our team and our team work to help our clients. This is great. Anything else that I forgot to ask you or anything else that you would like the listeners to know? I want to thank you personally, Alexandra, because after interacting with you and talking to you,

Aleksandra Zuraw (28:46.264)
throughout interactions, I have really came up with a lot of new ideas and it really crystallized what's really novel for our company and what's valuable. So I wanna just thank you for bringing all of this. This is so nice. Thank you so much. Thank you so much for being my guest today and for telling us what Histoiz is about and if...

one more thing, where can we find you online for more information about the company? Sure, you can go to our website, historist.com, H-I-S-T-O-W-I-Z.com. You can also call us at 1-833-PATHLAB, P-A-T-H-L-A-V. Okay. I'll serve it perfectly. I'm going to put this in the show notes as well.

Okay, then thank you so much and have a great evening. It was great talking to you. Thank you so much for time and thanks for entertaining me. Thank you for listening. For more great digital pathology resources, visit the Digital Pathology Consulting website and subscribe to our newsletter on digitalpathologyconsulting.com. After subscribing, you will get access to the free annotation guidelines

which will help you annotate slides consistently in all your digital pathology projects. Talk to you in the next episode!