A study published in Radiology, a journal of the Radiological Society of North America, shows that an artificial intelligence (AI) tool is highly efficient at detecting pancreatic cancer on CT (RSNA).
Among cancers, pancreatic cancer has the lowest five-year survival rate. By 2030, it is anticipated to rank as the second most common cancer-related death in the US. The outlook gets much worse once the tumor is bigger than 2 centimeters, so the best way to improve the grim outlook is to find the tumor early.
The main imaging technique for finding pancreatic cancer is CT, but it only detects about 40% of tumors that are smaller than 2 centimeters. A powerful tool is urgently required to assist radiologists in improving pancreatic cancer detection.
An AI system called deep learning is being studied by researchers in Taiwan as a computer-aided detection (CAD) tool for the detection of pancreatic cancer. They previously demonstrated that the tool could successfully differentiate between noncancerous and cancerous pancreatic tissue. However, that study relied on radiologists manually segmenting the pancreas on imaging, which is a time-consuming process. In the latest study, the pancreas was automatically recognized by the AI tool. This is a big step forward because the pancreas is close to many organs and structures and comes in many sizes and shapes.
The researchers used 546 pancreatic cancer patients and 733 control participants to make up the internal test group that the researchers used to develop the tool. In the internal test set, the tool’s sensitivity and specificity were 90% and 96%, respectively.
Following validation, 1,473 distinct CT exams from institutions across Taiwan were used. When separating pancreatic cancer from controls in that set, the tool had a 90% sensitivity and 93% specificity. It had a 75% sensitivity for detecting pancreatic cancers less than 2 centimeters in size.
According to study senior author Weichung Wang, Ph.D., a professor at National Taiwan University and the head of the university’s MeDA Lab, “the performance of the deep learning tool seemed on par with that of radiologists.” In this study, “the sensitivity of the deep learning computer-aided detection tool for pancreatic cancer was similar to that of radiologists in a tertiary referral center, regardless of the size and stage of the tumor.”
Dr. Wang claimed that the CAD tool has the capacity to offer clinicians a wealth of data. It might serve to expedite the radiologist’s interpretation by indicating the area of suspicion.
The study’s co-senior author, from National Taiwan University and National Taiwan University Hospital, said that the CAD tool “may serve as a supplement for radiologists to enhance the detection of pancreatic cancer.”
Future research is being planned by the scientists. They want to focus in particular on how the tool performs in more ethnically diverse populations. They want to see how it performs going forward in actual clinical settings because the current study was retrospective.
 tool is highly effective at detecting pancreatic cancer on CT, according to a study published in Radiology){kind=link}