AI Tool Rapidly Analyzes Gene Activities in Medical Images to Highlight Hidden Cancers

A novel artificial intelligence (AI) tool, designed to interpret medical images with exceptional clarity, is set to revolutionize the way clinicians approach disease diagnosis and image analysis.

This advanced tool, named iStar (Inferring Super-Resolution Tissue Architecture), was developed by researchers at the Perelman School of Medicine at the University of Pennsylvania (Philadelphia, PA, USA). It can assist healthcare professionals in diagnosing and treating cancers that might otherwise remain undetected. iStar offers an in-depth view of individual cells and a broader look at the full range of human gene activity, potentially revealing cancer cells that were nearly invisible earlier. This tool could play a crucial role in confirming whether cancer surgeries have fully removed malignancies and provide automatic annotations for microscopic images, marking a significant leap toward molecular-level disease diagnosis. One of the standout capabilities of iStar is its automatic identification of crucial anti-tumor immune formations known as "tertiary lymphoid structures," which are indicators of a patient's survival prospects and their likely response to immunotherapy. This precision makes iStar a powerful tool for selecting the right patients for immunotherapy treatments.

Image: A new AI tool brings precision pathology for cancer and beyond into quicker, sharper focus (Photo courtesy of 123RF)

The researchers put iStar to the test across various cancer types, including breast, prostate, kidney, and colorectal cancers, alongside healthy tissue samples. In these trials, iStar automatically detected tumor and cancer cells that were challenging to spot with the naked eye. With iStar as an additional support layer, clinicians might soon be able to diagnose more elusive cancers effectively. Moreover, iStar operates at a remarkably fast pace compared to similar AI tools. In a trial with a breast cancer dataset, iStar completed its analysis in a mere nine minutes, whereas the closest competing AI tool took over 32 hours to deliver a comparable analysis. This makes iStar an astounding 213 times faster, offering a significant advantage in time-sensitive clinical environments.

“The power of iStar stems from its advanced techniques, which mirror, in reverse, how a pathologist would study a tissue sample,” explained Mingyao Li, Ph.D., a professor of Biostatistics and Digital Pathology. “Just as a pathologist identifies broader regions and then zooms in on detailed cellular structures, iStar can capture the overarching tissue structures and also focus on the minutiae in a tissue image.”

Related Links:
Perelman School of Medicine