139: Why AI Won't Replace Pathologists and How It Can Help Them Thrive

Why are so many pathologists still afraid of going digital? 

In this USCAP special episode sponsored by Muse Microscopy, I talk with Dr. Sarah Dry, Chair of Pathology and Laboratory Medicine at UCLA, about real-world adoption, AI fear, and how change is best managed when it's people-led.

From her early digital research lab in 2007 to pioneering innovative workflows at UCLA today, Dr. Dry knows how direct-to-digital imaging and AI can enhance, not replace, our work.

🧠 Key Takeaways:

• [00:01:00] Fear of change and pathologist pushback
• [00:03:00] Direct-to-digital imaging & frozen sectios
• [00:05:00] Redefining histotech roles amid tech shortages
• [00:08:00] CAP standards, tissue storage & digital-first futures
• [00:12:00] Specialty-specific barriers to adoption
• [00:15:00] Why AI is a pathologist extender, not a threat
• [00:18:00] What AI can automate—cell counts, GI screenings & reports
• [00:20:00] Managing change and getting early adopters onboard
  
  

🎧 Tune in to hear how digital pathology becomes scalable, without losing the people at its core.

#DigitalPathology #AIinPathology #ChangeManagement #PathologistLeadership #WomenInSTEM

Transcript

Why AI won't Replace Pathologists, But It Can Help Them Thrive with Sarah Dry

Introduction: The Fear of Change

Sarah: [00:00:00] Most of us really don't like change. I think it's also scary and I think particularly, in a world where many people feel like they don't have a lot of control over other things in their environment. Once you start making changes to their work environment, which is an area where they do feel they have a lot of control, that can be even scarier.

Early Adoption in Digital Pathology

Sarah: One of the things that's really key is to find other early adopters.

Aleks: It's gonna help, it's gonna change workflows, and I'm obviously talking about the positive things. It's gonna improve efficiency, but you are right. A lot of these questions are fear-based. That comes back to change management and fear of change.

Sarah: I also think that a lot of pathologists are scared that this is gonna replace them. Digital pathology and AI will replace them. I've been around pathology long enough to have seen multiple different innovations that were thought to. They were gonna replace pathologists and we've not been replaced yet.

And now it's digital and it's never gonna replace us. It's gonna make us much more efficient and accurate.

Special Episode Introduction

Aleks: This episode is one of our special USCAP episodes that were all sponsored [00:01:00] by Muse Microscopy. So let's dive into it. 

Welcome my digital Pathology, trailblazers, another special episode from USCAP

Meet Dr. Sarah Dry

Aleks: And today my guest is Dr. Sarah Dry. She's a professor and Chair Pathology at UCLA. And she's actually the boss of one of our panelists from yesterday, Dr. Rao who you have seen in other podcast. And welcome Dr. Dry… 

Sarah: Thank you 

Aleks: …to the show. Let's talk about you first. Sure. Who are you? What's your background? And basically. What is your attitude towards digital pathology?

Dr. Dry's Journey with Digital Pathology

Aleks: I already know it's a positive one, but let the digital pathology trailblazers know a little bit more about you.

Yeah. It's so nice to meet you and to speak to you today on this podcast. So I am a professor of pathology at UCLA. I'm a bone and soft tissue in GI pathologist and I have, I consider myself an early adopter for technologies in general. So I was one of the first people [00:02:00] nationally to introduce digital pathology in the research setting.

Back in 2007 when I was running one of our research core labs at UCLA, so we introduced digital slide scanning 

Sarah: Yes. Yeah, that was when Aerio really brought out their first system and everybody was wondering what this was going to turn into, and obviously now it's become part of our routine care of patients and our routine pathology practice.

So I'm really excited about your system though, because. I think that we will never fully adopt digital pathology until we get rid of slides. And that's exactly what your system is going to be the first step towards.

Muse Technology and Its Impact

Aleks: Yeah, so the Muse technology, the direct-to-digital, I think it addresses the main pushback, main pain point that people main, like problem that people have with digital pathology because you have the early adopters, you have the enthusiast and they will make everything to make it happen. And I'd belong to that group as well. But you have the normal people who just wanna do their job, [00:03:00] wanna practice pathology, and if it means adding a machine on top of what they're already doing pretty well in an analog way.

And then there's this pushback and MUSE technology is taking this away because they image directed digital and that what the panel yesterday was about. When Dr. Rao was one of the panelists, he presented the research that they did in China.

What does his, where does his research and what he does with Muse and other things fit into the bigger picture at your institution?

Clinical Innovations at UCLA

Sarah: Dr. Rao is our vice chair for clinical innovation. So he's always on the lookout for new technologies that are going to be revolutionizing our field, and obviously, Muse is one of these, I like to give him a lot of freedom to explore these areas of interest, because…

Aleks: I think he's lucky to have a bus like that.

Sarah: I don't think he's lucky. He's very productive. [00:04:00] And he… So when I

Aleks: Yes, I heard like how much data they gathered in three days, I'm like, okay. Yeah. I would my person also go and get so much data.

Sarah: Yeah. And he has a long history of doing this. At UCLA, So…

Aleks: Okay. 

Sarah: I think that when people are themselves passionate about projects, sometimes the best thing you can do as a chair is to stay out of their way and let them continue on.

And really what's happened here. Rao has been, as, instrumental in identifying the partners in order to, provide the tissues that, that you need in order to do these clinical trials. And it's just really worked out very well.

Aleks: We had Dr. Zo on the panel as well. His partner is from China.

Sarah: Yeah.

Aleks: She had a cool piece of tech because she had glasses that would provide translation from the panel as well. 

Sarah: Oh, really? 

Aleks: Yes. And everybody was excited about the glasses, but such an amazing piece of equipment to exchange information now. The language barrier is not a barrier anymore.

Okay. We're gonna take the analog. [00:05:00] Hopefully, soon out of the way, where do you see this technology fitting into the workforce, and how do you see the deployment? Is it gonna replace the traditional way? Is it gonna complement the traditional way? And obviously Muse has a, like a very specific use case that they're working right now on.

But. Once we take the leap when it's ready. Once it's ready, where do you see it changing? Pathology, changing the way we practice, changing the way we provide care in general for basically everybody. For the pathologist, histotechnologists.

And everybody involved in these pathology-driven diagnostic decisions.

Sarah: The way that I see it is obviously it's gonna have to be a stepwise fashion as with anything. And I think the frozen section sphere is the first place that makes the most sense. To start, probably. But I do eventually see that this technology will replace traditional histology.

I think that rather than, spending all the money that we're currently spending [00:06:00] on chemicals, to preserve these tissues and then storing them in paraffin blocks and in slides that we have to then keep on site for years and years, I think that we're going just have a piece of tissue that we're gonna scan.

We're gonna. Perhaps freeze those or find some other way to store those. 

Aleks: That was a question that came up with the panel because obviously, there are requirements for how long you have to store the tissues. Which tissues like, depending on what diagnosis it is. Freezing did not come up as an answer, but yeah, totally.

Sarah: I think that could be, in the same way that we've developed things like RNA later, which preserves tissues for a long time, even though it may be that we do that it. It's possible that we're going to continue to paraffin-embed them, but that whole process affects the tissue quite a bit.

So I actually wouldn't be surprised if we just go directly from scanning to freezing and, in keeping a certain number of or maybe a weight, certain weight of tissue. Or a certain size of tissue, by, I think CAP standards [00:07:00] we'll have to change for all this, obviously. 

Aleks: Yes.

Sarah: But then, we'll keep those tissues for subsequent testing. But also, as certainly digital and AI is allowing us to identify a lot of features on histology that we as humans can't identify. And perhaps in the future, obviating the need for a lot of this molecular testing that we do.

So it may be that in the future we don't even need to keep the tissue because once we scan it we're gonna have all the information that we need. Yeah. And I truly believe that. This is just the first step, what we're doing with the tissue in, in the cassette that you've created.

I think at some point we will just scan the entire tissue.

Aleks: Yeah. Tell me how you see that. How what's your vision for this kind of workflow?

Sarah: I think that's super exciting because if you think about the way that we function as pathologists right now, it's not very precise. We like to, we try to be as precise as possible.

But we are using our eyes, which are not very. [00:08:00] But they're not able to discriminate microscopic foci of tumor at the margins. And so we choose the section that we think is gonna be closest approach to margin, for instance. And, that's a couple millimeters thick, and then we randomly take a slide out of that's four microns thick.

And that's where we make our decision on as to whether that margin is positive. But imagine if we could instead use your technology and, the computer's able to evaluate it and say, this is the closest approach to margin. Over a four or five or six centimeter area. This is the closest approach to margin.

Then we would know for sure if that margin's positive or negative. So to me, that's the most exciting aspect of it, is that, eventually you'll put the tissue in 10 minutes, you'll have representative slides that the pathologist will look at, and we'll determine what the tumor is, whether the margins are positive, whether there's LVI, whether the lymph nodes are positive, and you'll move on to the next case.

Aleks: That's so cool. So you are an early adopter, like you said. And [00:09:00] like we said, both said the change always happens.

Challenges in Change Management

Aleks: Stepwise, what is your experience and approach to change management at your institution? Like, how do we deploy these things? Because you already did that for the glass-based digital pathology, and let's say this is gonna be the next step.

What was your approach before? Would you change it? How do you navigate change management for new technologies in an organization? 

Sarah: Great question.

Change management is always difficult because people don't like change. Most of us really don't like change, so …

Aleks: Yeah, you have to learn so much, put so much effort.

Finally, you're comfortable, and they tell you to change.

Yeah. I think it's also scary and I think particularly, in a world where, many people feel like they, they don't have a lot of control over other things in their environment. Once you start making changes to their work environment, which is an area where they do feel they have a lot of control, that can be even scarier.

So I think, one of the things that's really key is to find other early adopters [00:10:00] within your organization. Or people who are at least willing to consider it. And because once you get those people on board, if they're excited, if they see positive changes, they start talking to their colleagues.

And then naturally other people will approach you and say, Hey, can we try it in my subspecialty area? And then once you get that momentum going. Then it's unstoppable. But yeah, it's never a good idea to tell people they're gonna have to do something that doesn't go over well.

Aleks: Is there a point where you, like you have those that lag behind and don't want to change and you wait for this push when it's mandated?

Sarah: Sure. 

Aleks: In the change cycle?

Sarah: Yeah. Sometimes you do get to that point where you just have to tell people that we're moving to this. But. The it's better to try to make people, or not make people, it's better for people to naturally come to that decision themselves. It's also important to understand what the concerns and pain points are, because sometimes there are legitimate concerns that, I may not recognize because I'm not a subspecialist in that area.

And there are sometimes small tweaks that can be made in order to [00:11:00] adapt to those concerns. So…

Aleks: Definitely it's a good point, because and I'm a veterinary pathologist, so I do not. And not even a diagnostic one. I work for drug development. So my workflow is not a diagnostic workflow, it's a study-based, cohort-based workflow.

So I don't work with cases, I work with groups of animals. 

Sarah: Yeah. And to your point, when digital pathology took off, it took off first in the diagnostic and clinical space, and then the vendors would come to the pharma industry and would suggest these solutions. For Toxicologic pathologists and very few understood the differences between workflows even now.

And there are like a handful of companies that decided to go and support these pathologists. Only a few understood the workflows, so that's a super important point. 

Sarah: Yeah. We have [00:12:00] at UCLA we've developed with in-house, a software platform for the tumor board. That we use with digital slides.

And my service, the bone and Soft tissue service, was one of the early adopters. And in the course of talking with some other services, I've realized that some of those services just can't adopt this technology as easily because of the way that their clinicians submit cases to tumor board or the types of information that's needed from the pathologist to the tumor board.

That's an all, another example of a system that to me, worked so beautifully 

Aleks: Exactly. 

Sarah: And made my workflow faster and more efficient. And another surface would actually be worse.

Aleks: And I always say about digital pathology, that this is a multidisciplinary field. And then I take like pathologists on one part of the field. But then what we are talking about, there are subspecialties, sub-workflows. Yes. Within pathology that will influence the excitement about the technology, the adoption that. People from outside of the pathology group. So other [00:13:00] stakeholders would be IT software development, I don't know who else is there.

The image analysis, AI teams, right? Multidisciplinary teams of different backgrounds. And then we are this pathology group, but then within the group, it's not homogeneous. So, figuring out the little tweaks, how to make it work. It's the same thing showing a slide, but showing it to a toxicologic pathologist is totally different than showing it to a diagnostic pathologist.

Yeah. And then depending on specialty, you also need different hardware capabilities. So something I did not think about until I heard about it was so that first I was surprised, oh, why does it the first time I talked to a guest who was doing this transition. And he said, “Oh, first this specialty went and then that specialty, but this specialty.

And he was, that was gin to urinary or kidney pathologist. I'm like, why? Oh, because they need to do the polarization and we don't have a tube for polarizing. 

Sarah: Yeah. [00:14:00]

Aleks: I'm like, of course. Yeah. So I don't do it in a routinely, so I didn't think about it. And now there are scanners that do that as well. 

Sarah: Yeah.

Aleks: But that's like a little thing. That, like this one thing that is different that this specialty is doing from everybody else that basically excludes them from being adopters of the technology. So yeah, a lot of analysis of different workflows.

Sarah: Yeah. I also think that a lot of pathologists are scared that this is gonna replace them.

Digital pathology and AI really place them. And I've been around pathology long enough to have seen multiple different innovations that were thought to. They were gonna replace pathologists and we've not been replaced yet. So when I was first starting my training, it was immunohistochemistry, right?

And then it was molecular. And now it's digital and it's never gonna replace us. It's gonna make us much more efficient and accurate. So I think

Aleks: It's gonna help, it's gonna change workflows. It's gonna, and I'm obviously talking about the positive things, it's gonna improve efficiency. [00:15:00] But you're right, a lot of these questions are fear-based.

That comes back to change management and fear of change. The same discussion when I was promoting the Muse panel on social media was from the histotechnologist side. 

Sarah: Yes. 

Aleks: So yeah, they were basically like very protective of the specialty, asking, oh, is it gonna replace us? You do not need any education.

And there's this discussion parallel to the same discussion that we're having with AI. What's gonna happen? And my take on that, those technologies, whichever that is, has a fantastic potential. But the application is a fraction of the potential that the technology has, and it's usually bumpy, so it still requires the subject matter expert to guide it.

As long as it makes your life easier [00:16:00] and not more difficult, then I'm an enthusiast of that. And, but what I wanted to say, in addition to the tech and specifications, the technicians, Oh yeah. Something that I learned from Rao in a previous interview that I did with him at CAP and he was talking about it, that there is actually a shortage of histotechnicians.

Sarah: Yes. 

Aleks: So, how are you gonna, and because of the shortage, there are less places to educate them. So there is no like mitigation strategy to educate more people to become technicians. And the job still needs to be done. So, how are we gonna do it? We have to solve it with technology. We have to solve it somehow.

Yeah. And Muse would be a candidate of a technology that could solve it. Then you're gonna have new people who will have to learn this technology. 

Sarah: And I think, the job of Histologist will, histotechnologists will evolve. And so there'll be new activities that they do. So I, I think all of this, and it's gonna happen over a long period of time. [00:17:00]

None of this is gonna happen rapidly. We'll have time to adopt to it.

Aleks: It's gonna be changing in parallel with the whole, like different way of doing pathology.

I don't know how much their job changed with immunochemistry with molecular, but definitely the workflows changed because you have to treat tissue differently.

So they had to learn new things and so did we have to learn new things.

The Role of AI in Pathology

Aleks: And one question that I wanna ask you is about AI. Of course, we cannot finish this conversation without talking about AI. 

What is your current experience? Are you guys implementing anything? Are you guys working on anything?

What are your plans? How should AI support you? How is AI currently supporting you in your pathology work and your team?

Sarah: So right now we don't have AI supporting us, but we are actively looking into it. I think that there are a lot of ways that it can and should help us again. Okay. To me it's a physician extender, it's a pathologist [00:18:00] extender and a pathologist assistant…

Aleks:  I see. The same way. 

Sarah: For instance, our hematopathologists spend a lot of time right now counting cells in, peripheral smears and bone marrow biopsies and to me that's something that AI can do for them, and they can confirm, but. I think that they spend a lot of time doing that, and that, to me that's an area where they can be more efficient.

We could use, we want them to be functioning at the top of their skillset. 

Aleks: Okay. 

Sarah: And I would argue that counting cells is not the top of their cell. Their skillset. No counting. And particularly now that they have to integrate molecular data and clinical data into their final diagnosis.

And a lot of it is also, the communications are more complicated with clinicians because the information that we're providing clinicians is more. Complicated and nuanced. I would rather them spending their energy in those areas than counting cells. And, same with other areas of pathology.

I think that AI is going to be making routine diagnoses for us and also it's going to be providing draft [00:19:00] reports for us, which is gonna save a lot of time. For a GI pathologist, a lot of you know the colon biopsies are preneoplastic lesions, tubular adenomas, hyperplastic polyps. Wouldn't it be wonderful if we had a scanner that could scan the slide and put up…

Aleks:  And screen them

Sarah: …and create the report, and you still have to check it.

But it's gonna be a lot faster.

Aleks: So you guys are looking into it? 

Sarah: We're looking into all of those things. And so right now there there's some FDA-approved AI products out there that we're looking at. And again, that's  part of Dr. Rao's job is vice chair for clinical innovation, is just…

Aleks:  What's out there.

Sarah: And we also have some of our clinical informatics folks are also. On the market and, making sure that they know what's out there. Yeah. What, I hope that we are gonna be adopting some of those in the very near future.

Conclusion and Final Thoughts

Aleks: Thank you so much. 

Sarah: Thank you.

Aleks: So much for, this is really fun joining me. 

Sarah: Yeah. 

Aleks: And thank you so much also for I love to showcase pathology, female leaders. Oh. That's like a kind of side benefit of of our conversation. Yes. I'm gonna highlight you in the International Women's Day. Short video that I always have.

Sarah: Wonderful. Thank you. 

Aleks: Thank you.

Sarah: Thanks so much, Alex. It was really nice talking with you.