Challenges and Opportunities: A Special Edition on AI for Pathology

July 14, 2026

CAP AI Studio

Transcript

Nick Lanyi 0:02

The age of artificial intelligence. What does that mean for pathology?

Dr. Rama Gullapalli 0:07

What we are talking about is changing our entire thought pattern, actively being influenced on a day-to-day basis. And that is not something that we can, you know, um trivially implement as a proficiency test. It has to be a very thoughtful process happening at the college level, at the institutional level, and at a personal level as well as a professional pathologist.

Nick Lanyi 0:38

This is Path News Network, daily edition from the College of American Pathologists. I'm Nick Lanyi with a special edition of the podcast. It's Tuesday, July 14th. Since Chat GPT captured the public imagination in November 2022, AI has been the next big thing, dominating headlines and public discussions. Of course, pathologists and radiologists have used machine learning tools to analyze images and data for many years. That AI would be the future of pathology was clear enough that the CAP launched its AI committee a full seven years ago. But over the past year or two, progress has accelerated. Pathologists are increasingly adopting AI tools for clinical decision support, personalized medicine, automation, reporting, research, and other applications. And while digitization rates remain low at 10 to 15 percent of laboratories, growing adoption of digital slides will only fuel the AI fire. And it's a lot to get used to. Here's Dr. Matthew Hanna, Vice Chair of Pathology Information at the University of Pittsburgh Medical Center and chair of the CAP's AI committee.

Dr. Matthew Hanna 1:50

With the technology uh development uh happening so rapidly, it is a constant uh learning curve of everything that's out there, how the field is evolving, how the technology is evolving. And so what I think has been challenging is how do you wrap standards with you know standard standard practices or uh checklist items with some of these.

Nick Lanyi 2:16

The CAP is responding in several ways, including working with peer organizations on digital pathology interoperability and standards, as well as ramping up member support on how to use AI in the laboratory.

Dr. Matthew Hanna 2:29

We are at the point where we will be able to actually start offering uh you know uh an implementation guide for pathologists on how to actually use these tools. So I think we put a you know a very big effort into the education. We've had webinars and podcasts and um some early manuscripts publications about what the tools are, some education around there, um, and then how to actually go about validating these tools. And then now are coming out with sort of targeted practical approaches on how to actually plug these into your lab and uh are starting the process to hopefully formally provide guidelines to um to the to the members of the college and to the community.

Nick Lanyi 3:18

The CAP AI implementation series is scheduled to be released by the end of the year. It will include guidance on validation, deployment, and monitoring of AI tools, recommendations on reporting when using AI applications, and more. The CAP also offers members a unique way to try out a range of different AI applications for free. The CAP's AI Studio, launched in 2025, offers a range of products for many top vendors. You can search by test, then find the appropriate products and play around with them using sample case slides. Dr. Anil Parwani, Chair of the Department of Pathology at Ohio State University and a member of the CAP's AI committee, encourages members to check it out.

Dr. Anil Parwani 4:09

This AI Studio is a phenomenal idea because it brings those tools from many, many different vendors right in front of the pathologist where they can take some test data and use the tool and see the interface, see how it processes, how it does it give the results in the right format.

Nick Lanyi 4:31

The CAP plans to continue expanding the vendors and products available in the studio so that members can compare applications before their laboratory buys them.

Dr. Anil Parwani 4:40

You know, it's almost like test driving a car. You know, when you want to buy the car, you want to see how it feels. Does it do its job? Uh are you happy when you sit in this car?

Nick Lanyi 4:51

You can find the CAP's AI Studio at ai studio.cap.org. Dr. Rama Gullapalli, Associate Professor of Pathology, Chemical, and Bioengineering at the University of New Mexico, is also a member of the CAP's AI committee. He has focused on ethical issues raised by AI.

Dr. Rama Gullapalli 5:15

So AI technologies are very powerful. Um at the same time, if they are adopted without much thought and concern, there is a real danger of pathologists losing their core value in terms of healthcare and the skill sets they bring to the table. And this would be this topic known as de-skilling and how is that going to impact the practice of pathology is one of the things that interests me. There is another entity known as automation bias, which is essentially once AI is in place, uh humans tend to kind of overly rely on the output from AI to the to the exclusion of their own hard-earned skills.

Nick Lanyi 6:04

Placing too much trust in AI is a problem for many professions. But Dr. Gullapalli thinks pathology is especially vulnerable.

Dr. Rama Gullapalli 6:11

Especially in a time pressure-driven specialty like pathology, where, for example, if you're in a community pathology practice and you have been assigned a tool that this is something that you can rely on without providing a proper description of what the hazards are, over time, what will happen is you will kind of not second guess or just adapt what it has to kind of offer you without concern for what times it can go wrong. And again, what we all need to remember is that AI is not infallible. It will make mistakes. And if one gets sidetracked due to issues like automation bias, that becomes a hazard for patient care in the long run.

Nick Lanyi 7:00

The CAP can play a role in addressing these risks by promoting AI governance policies and other guidelines, Dr. Gullapalli said.

Dr. Rama Gullapalli 7:08

Focusing on these biases as a part of the um accreditation standards. We can think about designing peer review programs that you know test the pathologists against the AI, AI against pathologists, and then peer groups against peer groups in terms of AI usage or so forth.

Nick Lanyi 7:29

But ultimately, Dr. Gullapalli said, AI will require a different philosophical approach.

Dr. Rama Gullapalli 7:36

This is just not a tool. This is just not an instrument, unlike your routine stuff. Over there, you know, there are a lot of safety guards built in, there are proficiency tests that can handle it. But with AI, what we are talking about is changing our entire thought pattern, um, actively being influenced on a day-to-day basis. And that is not something that we can, you know, um trivially implement as a proficiency test. It has to be a very thoughtful process happening at the college level, at the institutional level, and at a personal level as well as a professional pathologist.

Nick Lanyi 8:17

Despite the many challenges, he's optimistic about AI's impact on pathology.

Dr. Rama Gullapalli 8:23

Where AI, I think, uh is not as good, will probably not get there anytime soon, is its ability to merge and integrate uh diverse streams of information, which is what I'm trying to say is that um we pathologists tend to forget that we are physicians, which means we have a strong basis of clinical uh education and training. We need to increasingly bring that back into the practice. Your uh ability to diagnose a patient should not solely rest on the foundation of being able to detect a pattern or a disease or a cancer. What value addition you are creating by your ability as a human to integrate diverse streams of information from the clinical element to the molecular element to the imaging element in crafting a diagnostic narrative that not only diagnoses the patient but also prognosticates how they are going to respond, what therapies are going to be useful. So, this is where, as a molecular pathologist, I feel very, very uh excited about the future of pathology.

Nick Lanyi 9:38

Dr. Parwani also thinks AI will ultimately help pathologists help patients.

Dr. Anil Parwani 9:44

Uh so I think the biggest benefit right now is the um if you look around globally, there's a shortage of pathologists. There is also an increased number of cases per pathologist, and the complexity of the cases is going up. So, what I used to know when I was a resident in my training, that has amplified. You know, there's a you're basically learning uh from the fire hydrant every day. There is so much information out there. So uh on a daily basis, it's hard to keep up with it. So AI can help us synthesize the information, find bring the relevant information, check images for quality. Quality is a big thing.

Nick Lanyi 10:26

We'll give Dr. Hanna the last word.

Dr. Matthew Hanna 10:30

I'll steal a quote from um uh Dr. Cardona, who's on another CAP committee. But you know, I think pathologies, you know, we we've historically held this title as being the doctor's doctor, and I think we are adding another label of trying to be the patient's partner along this journey where we are using a lot of these tools and and being more of an interface within uh the patient care workflows.

Nick Lanyi 10:55

Well, that's all for today. See the show notes for more information, and feel free to share this special edition on social media. We'll be back Wednesday at 5 a.m. Eastern for another episode of the Daily Edition. I'm Nick Lanyi. Have a wonderful day.