AI in pathology: excitement vs. fear
Donald S. Karcher, MD
September 2024—The use of artificial intelligence in medicine is spreading rapidly, but it’s not entirely new. Tools that incorporate AI have been approved by the FDA since the mid-1990s, mostly for use in radiology and cardiology. Like all of our colleagues in medicine, we pathologists have been hearing a lot about AI lately. Some of us are excited about it, and most of us are familiar with the surge of hype claiming that AI will change everything about how we practice.
But what AI is, and how exactly it might prove useful in pathology, is still not clear to many of us. It feels new and mysterious, triggering a lot of anxiety about what it means for our profession, our practices, and our patients. I’ve heard a lot of fear that AI may eventually even replace us.
This reminds me of the fears we felt when immunohistochemical stains and other proteomic methods were first being developed, and again when molecular genomic analysis was introduced into pathology. How many of us remember claims that these technologies would make microscopy somehow less important, that one day we might not even need microscopes? All these years later, we still have our microscopes, or computer workstations to view digital slide images, and while IHC and genomic analysis haven’t replaced pathologists, we have found them to be valuable tools in delivering better answers for our patients.
Like many other “disruptive” technologies, AI is just another tool that will allow us to make better and more actionable diagnoses—nothing more and nothing less. It will not take our place and it will not eliminate the need for our expertise.
But we should be prepared. There are loads of pathology AI tools that are on the verge of being FDA approved, and I expect they’ll start coming out by the dozens in the next few years. Image analysis tools will be quite common (there are already FDA-approved AI programs for image analysis in pathology) and so too will be products to analyze “big data” across many clinical sources, including laboratory data. We will have to know how to assess them, select the best ones for our practices, and implement them.
So let’s look at the different types of AI offerings. Assistive AI is meant to assist pathologists with basic tasks. Current image analysis tools often fall under this category; they can locate cells and patterns of interest that the pathologist can then examine and use to make a diagnosis. Augmentative AI tools go beyond simple assistance, enhancing what pathologists can do on their own. These programs might be able to recognize and analyze patterns that are not readily identifiable by eye or with existing pathology tools. The most advanced AI tools are autonomous, capable of coming to their own conclusions without human involvement or interaction. With this model, the pathologist would review the AI’s conclusion and accept, question, or overrule it. Finally, the AI we hear about most often in mainstream media is generative AI, like ChatGPT. This model is emerging quickly in medicine with AI tools that can create text, images, and other outputs by analyzing data from the electronic health record and many other sources.
In all of these cases, it’s the combination of human and machine that leads to better outcomes, not the AI alone. Studies have shown that AI tools are very good at analyzing clinical and laboratory data, but they are not nearly as effective as the physician in understanding and incorporating clinical nuances in diagnosis and treatment, or in developing the clinical hunches that often lead to correct patient management decisions.
While AI tools will help us do our jobs, the use of AI in pathology will also involve many challenges, including quality management, regulatory oversight, cost, and payment. The CAP is already hard at work addressing these issues. For example, our Artificial Intelligence Committee is developing resources that will help members with the selection and implementation of AI tools in all aspects of pathology practice. The CAP Learning Division is also creating educational resources to help us understand and use AI.
Externally, our advocacy team is working with several government agencies on how pathology AI can be best regulated. The CAP’s Economic Affairs Committee is working with the American Medical Association CPT editorial panel and with the Centers for Medicare and Medicaid Services on how payment of pathologists and laboratories can be fairly structured for use of AI in pathology. Finally, the CAP is a founding partner of the Coalition for Health AI, or CHAI, a prominent national organization focused on the development of safe, effective, and ethical AI tools in medicine.
The bottom line for pathologists is this: We will need to embrace AI tools as another technology that will allow us to provide richer, deeper, and more actionable diagnoses and laboratory data that will result in better outcomes for our patients. There is nothing to fear. We have been here before.
Dr. Karcher welcomes communication from CAP members. Write to him at president@cap.org.