New AI Tool Reveals Hidden Genetic Signals in Routine H&E Slides

Pathologists worldwide rely on hematoxylin and eosin (H&E) slides to examine tissue architecture, yet these stains do not reveal the underlying molecular activity that often drives disease. As workloads rise and access to specialized molecular assays remains uneven, laboratories need tools that surface clinically relevant signals from routine specimens. Spatial biology approaches can expose complex cellular interactions within the tissue microenvironment that standard microscopy cannot capture. A new study now shows an artificial intelligence (AI) system can extract such information from everyday slides to support earlier and more precise diagnoses.

Researchers at QIMR Berghofer Medical Research Institute (Brisbane, Australia) have developed STimage, an AI screening tool that leverages spatial biology to predict hidden genetic markers and disease signatures directly from H&E-stained tissue. Built with machine learning and statistical algorithms, the system applies spatial analysis across the slide to generate biologically grounded diagnostic predictions. It also quantifies prediction certainty and surfaces the tissue and cellular features that contributed to the output, enabling transparent review by pathologists.

According to the study, STimage accurately predicted breast, skin, and kidney cancers, as well as the liver immune disease primary sclerosing cholangitis. The tool was described as reliable, low cost, and capable of rapidly generating results that pathologists could interpret. Researchers trained the model on de-identified datasets encompassing the aforementioned conditions.

The system produced accurate prognostic and treatment-response predictions, classifying patients as high or low risk of survival and as likely to have a complete or partial response to existing drugs. These prognostic and therapeutic features are at an early stage and are being further developed. In head-to-head comparisons noted by the team, STimage outperformed a small number of comparable tools while adding reliability and interpretability features.

The research is published in Nature Communications. Development was led within QIMR Berghofer’s National Center for Spatial Tissue and AI Research (NCSTAR). The team is expanding the range of detectable cancer types, increasing accuracy, and integrating additional datasets to identify rarer cancer cells at an early stage and key immune cell types that influence cancer progression and drug response. The next phase involves trials in pathology laboratories.

“It’s like giving pathologists the super-resolution vision of Superman or Superwoman to scan millions of invisible biomarkers in a tiny tissue sample to find the two or three that are showing signs of cancer. This capability is critical for earlier detection, more precise diagnosis, and better-informed treatment decisions,” said Associate Professor Quan Nguyen, who led development of the tool with QIMR Berghofer’s National Center for Spatial Tissue and AI Research (NCSTAR).

“The STimage tool does not replace the experience and expertise of pathologists. Rather, it assists them in their important and technically challenging work, by providing extra information about cell types and genetic activity that they can't see with their own eyes,” said Associate Professor Nguyen.

Related Links
QIMR Berghofer Medical Research Institute