Researchers have created a powerful AI tool that can quickly and accurately identify subtypes of pancreatic ductal adenocarcinoma (PDAC), the most common type of pancreatic cancer. This tool uses deep learning to analyze routine pathology images, offering a faster and cheaper alternative to traditional molecular testing methods.
The study, published in The American Journal of Pathology, highlights how the AI model can classify PDAC into two key subtypes—basal-like and classical—using standard hematoxylin-eosin (H&E) stained slides. This approach is faster and more cost-effective than current DNA or RNA-based tests, which often face delays due to the complexity of processing tumor samples.
Pancreatic cancer is one of the deadliest cancers, recently overtaking breast cancer as the third leading cause of cancer deaths in North America. Early detection is crucial, but most cases are diagnosed at advanced stages, where treatment options are limited.
Dr. David Schaeffer, one of the study’s co-leaders, explained that this new method overcomes challenges posed by the limited tissue samples often available in pancreatic cancer cases. He said, “Our study shows that AI can help classify PDAC subtypes using simple, widely available slides, paving the way for better treatment decisions.”
The AI model was trained using images from cancer databases and tested on additional patient samples. It achieved a high accuracy rate—96.19% on the training data and 83.03% on the local dataset—demonstrating its reliability across different settings.
Dr. Ali Bashashati, another co-lead researcher, highlighted the tool’s ability to assist in diagnosing and triaging patients for further molecular testing. He said, “This technology allows us to identify key cancer subtypes at the time of diagnosis, making it a game-changer in personalized treatment.”
This breakthrough offers new hope for improving outcomes in pancreatic cancer by providing faster, more accessible, and cost-effective diagnostic options.
Source: webwire
