UChicago AI Tool Helps Diagnose Rare Thymic Tumors with High Accuracy

A team led by researchers at the University of Chicago has developed an artificial intelligence tool that helps pathologists accurately diagnose a rare group of cancers known as thymic epithelial tumors (TETs), according to a study published in Annals of Oncology. TETs arise from the thymus — a small gland in the upper chest — and affect just 2 to 3 people per million each year, making them extraordinarily difficult for most pathologists to classify correctly.

Senior author Dr. Marina Garassino, a UChicago Medicine oncologist who specializes in thoracic cancers, said previous research in Italy had found a discrepancy of about 40% in TET diagnoses at non-academic centers. Because TETs come in five main subtypes that look and behave very differently, misclassification can send patients down the wrong treatment path for aggressive cancers.

“TETs arise from the thymus — a small gland in the upper chest — and affect just 2 to 3 people per million each year, making them extraordinarily difficult for most pathologists to classify correctly.”

To close the gap, the team trained a deep-learning model on microscope images from 119 patients drawn from The Cancer Genome Atlas, where expert pathologists had confirmed subtypes. They then tested the tool on 112 cases from UChicago, again with expert-confirmed diagnoses. The model classified TET subtypes with high overall accuracy and correctly identified 100% of thymic carcinomas — the most aggressive subtype — outperforming non-expert pathologist diagnoses in the test set.

The tool is freely available and is not meant to replace pathologists but to support those without specialized training. The researchers are now validating it at larger scale with data from additional cancer centers in the United States and Europe, and plan to extend the algorithm to correct for differences in how slides are prepared and scanned between hospitals. Done well, that work could make the tool a practical aid in rural and community hospitals where patients with rare cancers often first receive their diagnoses.