Washington, Oct. 27 (SANA) An artificial intelligence system has shown promise in detecting early signs of pancreatic cancer more than a year before clinical diagnosis, potentially offering a new approach to identifying one of the deadliest forms of cancer.
The system, known as REDMOD, identified subtle changes in pancreatic tissue a median of 475 days before diagnosis, even when conventional computed tomography (CT) scans showed no obvious signs of the disease, according to a study reported by ScienceDaily.
Researchers analyzed CT scans from 219 patients who later developed pancreatic cancer and compared them with scans from 1,243 people who remained cancer-free during a follow-up period of up to three years, Sputnik reported Monday.
The research team included Sovanlal Mukherjee of the Mayo Clinic in the United States and specialists in radiology, artificial intelligence and medical imaging from other institutions.
REDMOD detected 73% of cancer cases, compared with 39% identified by radiologists reviewing the same scans, the study found.
For scans taken more than two years before diagnosis, the AI system achieved a detection rate of 68%, compared with 23% for radiologists.
The technology identifies subtle tissue patterns that may be difficult to detect through conventional image analysis. It can also automatically locate and outline the pancreas in CT scans, reducing the need for manual processing.
Researchers cautioned that further studies involving larger and more diverse patient populations, particularly people at high risk of pancreatic cancer, are needed before the technology can be introduced into routine clinical practice.
They also stressed that additional research must establish whether earlier detection using the system can improve treatment outcomes.
Pancreatic cancer remains difficult to diagnose in its early stages because symptoms are often absent or nonspecific, while early tissue changes can be hard to identify on standard imaging scans.
R.H/ADB
