AI in Pathology: Teaching Computers to Spot Cancer in a Slide

A pathologist's workday looks nothing like television. Most of it is quiet, precise work: a microscope, a stack of glass slides, and hours spent looking for small details that change a diagnosis. When a patient has a biopsy, the sample is stained, sliced thin, mounted on a slide, and examined for signs of cancer, infection, inflammation, or other disease. The report that comes out of that process can shape the entire treatment plan.

It is painstaking work, and there are not enough people trained to do it.

The global shortage of pathologists is a slow-moving crisis in the field. High-income countries are facing retirements that are outpacing training pipelines. Low-income countries often have only a handful of pathologists for millions of patients. The result is delayed diagnoses, which in oncology translates directly into delayed treatment and worse outcomes.

Digital pathology, which involves scanning slides at high resolution and working with images on a screen rather than through a physical microscope, has been gaining ground for several years. The transition to digital created the conditions for AI to enter the picture. A digital slide is, at its core, a very large image. And reading patterns in large images is something deep learning models do extremely well.

PathAI and Paige.AI are among the companies that have built AI systems for analyzing pathology slides. Paige received FDA clearance in 2021 for its prostate cancer detection tool, the first AI system cleared for primary diagnosis in pathology. Studies published in Nature Medicine showed that pathologists working with AI assistance caught more cancers than those working without it, particularly in subtle cases that a single reviewer might miss.

The AI systems are not diagnosing independently. They are highlighting regions of interest on a slide, essentially telling the pathologist where to look more carefully. For a pathologist reviewing fifty slides in a day, that kind of triage has real value. It reduces the chance that a small cluster of abnormal cells gets overlooked because of fatigue or volume.

What the technology does particularly well is consistency. Human pathologists are subject to fatigue and variation between reviewers. Studies have shown meaningful disagreement between experienced pathologists reading the same slides, particularly for cancers with complex grading systems like breast cancer and prostate cancer. An AI system, once trained, applies the same criteria every single time.

The version of this that makes the most sense is practical: AI helps with the volume problem by flagging routine cases, pointing out suspicious regions, and giving pathologists more time for the cases that need judgment and clinical context.