An artificial intelligence tool developed by the Mayo Clinic has shown the capability to detect pancreatic cancer on routine CT scans up to three years before clinical diagnosis, according to a study published in the journal Gut by the British Society of Gastroenterology. Led by Sovanlal Mukherjee from the Department of Radiology at the Mayo Clinic in Rochester, Minnesota, the study highlights the potential of AI to significantly improve early detection rates of a cancer that often proves fatal due to late diagnosis.
The AI system, known as REDMOD, focuses on what researchers term “imaging-occult” pancreatic cancer. This refers to cases where no visible tumors can be identified on scans, yet the disease is already manifesting through microscopic changes within pancreatic tissue that are undetectable by the human eye.
In testing, REDMOD was applied to 493 CT scans, employing a realistic patient ratio of approximately six healthy individuals for every pre-diagnostic case. The AI achieved a remarkable sensitivity of 73 percent in identifying pre-diagnostic scans, nearly doubling the detection rate of expert radiologists, who identified less than 39 percent. As the lead time extended beyond two years before diagnosis, REDMOD’s sensitivity was almost three times greater than that of human readers.
The system analyzed nearly 1,000 radiomic features extracted from each pancreas, narrowing them down to 40 key signals. Approximately 90 percent of these signals originated from wavelet-filtered images, a mathematical technique that reveals subtle textural disruptions in tissue often invisible to conventional imaging methods. Researchers believe these disruptions indicate early biological remodeling of the pancreas, occurring before any tumors are present.
To create its predictive model, REDMOD integrates three machine learning algorithms—logistic regression, random forest, and XGBoost—utilizing a soft-voting mechanism to arrive at a final classification. The model’s detection threshold is adjustable, allowing clinicians to balance sensitivity against false positives based on clinical context. Its precision rate of 36 percent already exceeds the 3 percent standard set by the UK’s National Institute for Health and Care Excellence for initial cancer referrals.
Longitudinal testing demonstrated a prediction consistency of 90 to 92 percent across repeat scans from the same patients. Furthermore, researchers confirmed that the model maintained its accuracy across two external datasets, including a public collection from the National Institutes of Health involving healthy volunteers, demonstrating robustness across various CT scanner brands and image quality settings.
Looking ahead, researchers are preparing to launch a prospective clinical trial named AI-PACED to evaluate the tool in real-world high-risk populations before it can be integrated into standard care practices. This development could mark a significant advancement in the early detection of pancreatic cancer, potentially improving patient outcomes by allowing for earlier and more effective intervention.
See also
AI Study Reveals Generated Faces Indistinguishable from Real Photos, Erodes Trust in Visual Media
Gen AI Revolutionizes Market Research, Transforming $140B Industry Dynamics
Researchers Unlock Light-Based AI Operations for Significant Energy Efficiency Gains
Tempus AI Reports $334M Earnings Surge, Unveils Lymphoma Research Partnership
Iaroslav Argunov Reveals Big Data Methodology Boosting Construction Profits by Billions
