AI Model Simultaneously Detects Multiple Genetic Colorectal Cancer Markers in Tissue Samples

Colorectal cancer is a complex disease influenced by multiple genetic alterations. Traditionally, studies and diagnostic tools have focused on predicting only one mutation at a time, overlooking the interplay of co-occurring mutations and shared morphological patterns. This limited approach restricts clinicians’ ability to fully understand tumor biology and hampers precision in patient stratification. Now, a new artificial intelligence (AI) method has been shown to simultaneously detect a wide range of genetic colorectal cancer markers directly from standard pathology slides.

The innovation was developed by researchers at Technische Universität Dresden (Dresden, Germany) through a multicenter collaboration involving institutions in Europe and the United States. The team created a novel “multi-target transformer model” that analyzes routinely stained histological tissue sections from colon cancer patients. The multicenter study evaluated nearly 2,000 digitized slides across seven independent cohorts, using both whole-slide images and supporting clinical, demographic, and lifestyle data.

image: Researchers Marco Gustav (right) and MD Nic G. Reitsam (left) discuss the study data (Photo courtesy of Anja Stübner/EKFZ)

The findings, published in The Lancet Digital Health, demonstrated that the new model matched and in some cases exceeded established single-target approaches for predicting biomarkers such as BRAF and RNF43 mutations, as well as microsatellite instability (MSI). Validation involved comparisons across independent patient cohorts, supported by expert pathological review. The study highlighted the frequent occurrence of mutations in MSI tumors, a key biomarker for identifying patients who may benefit from immunotherapy.

These results underscore how AI can accelerate diagnostic workflows and offer deeper insight into the relationship between molecular and morphological changes in colorectal cancer. In clinical practice, the tool could be applied as an effective pre-screening step, enabling more efficient selection of patients for molecular testing and guiding personalized treatment strategies. The research team now aims to extend the approach to other cancer types to explore its broader potential.

“Our research shows that AI models can significantly accelerate diagnostic workflows. At the same time, these methods provide new insights into the relationship between molecular and morphological changes in colorectal cancer,” said Jakob N. Kather, Professor of Clinical Artificial Intelligence at the EKFZ for Digital Health at TU Dresden and senior oncologist at the NCT/UCC of the University Hospital Carl Gustav Carus Dresden. “In the future, this technology could be used as an effective pre-screening tool to help clinicians select patients for further molecular testing and guide personalized treatment decisions.”

Related Links:
EKFZ Digital Health
Technische Universität Dresden