AI tools to finds pancreatic cancer early: study

Scientists in the United States have developed a new AI system that can detect pancreatic cancer much earlier than doctors. This tool, called REDMOD, was created by researchers at Mayo Clinic and University of Texas MD Anderson Cancer Center.

The study titled “Next-generation AI for visually occult pancreatic cancer detection in a low-prevalence setting with longitudinal stability and multi-institutional generalisability” was led by Sovanlal Mukherjee, along with Ajith Antony, Nandakumar G Patnam and Kamaxi H Trivedi.

According to the study published in BMJ journals, REDMOD studies CT scan images and looks for very small changes in the pancreas that are usually not visible to human eyes. These small changes can be early signs of cancer.

The AI detected cancer in nearly 75% of cases and, on average, found it around 16 months before doctors gave a formal diagnosis. This is almost double the detection rate compared to specialist doctors without AI support.

In some cases, the system even spotted signs more than two years earlier. The AI was trained on 969 CT scans and later tested on a different set of images. Out of 63 patients who were earlier declared healthy, REDMOD correctly identified 46 cases as suspicious.

However, the system is not fully accurate, cited the study. Out of 430 healthy people, 81 were wrongly flagged, meaning they may have been called for extra tests before being cleared. Still, experts say the tool shows strong potential. A radiologist from Mayo Clinic said the AI can detect cancer signs even when the pancreas looks normal.

Pancreatic cancer is expected to become one of the leading causes of cancer deaths in the US by 2030. Early detection can help save many lives.

Researchers said more testing is needed before using this AI in regular hospital practice.