Pancreatic cancer is one of the most dangerous and deadly forms of cancer in the world. It is often called a “silent killer” because it usually does not cause clear or obvious symptoms in the early stages.
When symptoms do appear, they can be very general, such as stomach discomfort, mild pain, loss of appetite, or unexplained weight loss. These signs are easy to ignore or may be mistaken for common health problems.
As a result, many people are only diagnosed when the cancer has already spread and treatment options are very limited. Sadly, only about one in ten patients survives for more than five years after diagnosis.
Another reason pancreatic cancer is so deadly is that it is very difficult to spot on medical scans. Even with modern CT scanners, small tumors in the pancreas can be hard to see, especially in the early stages.
The pancreas is located deep inside the body, behind other organs, which makes it more challenging for doctors to get clear images. Radiologists, who are doctors trained to read medical scans, do their best to identify signs of cancer, but even experienced specialists can sometimes miss small or unusual tumors.
Because of these challenges, a group of researchers decided to explore whether artificial intelligence, also known as AI, could help doctors detect pancreatic cancer more accurately. AI is a type of computer technology that can learn patterns by studying large amounts of data.
In recent years, AI has already been used in many areas of medicine, such as reading X-rays, detecting lung problems, and spotting signs of breast cancer. The research team, led by AI expert Henkjan Huisman and radiologist John Hermans, wanted to see if this technology could do the same for pancreatic cancer.
To test this idea, the researchers created a special and highly reliable set of medical scans. These scans were collected from nearly 400 patients in Western countries and were carefully checked by many international experts.
This collection of scans was kept confidential to make sure that it could be used as a fair and accurate test for AI systems. After preparing this data, the team invited developers from around the world to build and submit AI models that could try to detect pancreatic cancer in these scans.
More than 250 AI models were submitted for the challenge. Each model was tested on the same confidential set of scans and then compared with the performance of a group of trained radiologists. The results were surprising and encouraging.
The best AI models were more accurate than the average radiologist. In fact, the top AI system made 38 percent fewer false alarms, meaning it was less likely to say cancer was present when it was not. It also correctly identified cancer in 92 percent of the scans, while the radiologists achieved an average accuracy of 88 percent.
This does not mean that AI should replace doctors. Instead, it suggests that AI could become a very helpful tool to support them. Radiologists often work under great pressure and must review many scans every day.
An AI system could act like a second set of eyes, helping to highlight areas of concern and reduce the chances of human error. Over time, this could also reduce the workload and stress on medical staff.
The researchers are especially hopeful that AI could help detect pancreatic cancer earlier than is currently possible. Finding the disease at an early stage could make surgery and other treatments more effective and give patients a much better chance of survival. However, the team is also being very careful.
They understand that false positives, which are incorrect results suggesting cancer when none is present, can cause great fear and anxiety for patients and can place extra strain on the health care system. That is why they are continuing to train and test the AI on even more scans before it can be used in real hospitals.
When reviewing and analysing the findings of this study, it is clear that this research represents an important step forward. The creation of a reliable and carefully checked benchmark made it possible to compare AI fairly with human experts.
The fact that some AI models performed better than trained radiologists is a strong sign of the technology’s potential. At the same time, the researchers are right to move slowly and carefully, because medical decisions must always put patient safety first.
Overall, this study shines a small but important light of hope in a very dark area of medicine. Pancreatic cancer has long had very few encouraging developments, and any progress in early detection could make a meaningful difference. While AI is not a cure and cannot solve every problem, it may become a powerful partner for doctors in the fight against this deadly disease.
If you care about cancer, please read studies that a low-carb diet could increase overall cancer risk, and vitamin D supplements could strongly reduce cancer death.
For more information about health, please see recent studies about how drinking milk affects the risks of heart disease and cancer and results showing higher intake of dairy foods linked to higher prostate cancer risk.
The study is published in The Lancet Oncology.
Copyright © 2025 Knowridge Science Report. All rights reserved.
