Using AI to detect pancreatic cancer

Pancreatic cancer is the third leading cause of cancer death in the United States. That’s due in part to the limited testing available for early detection.

Now, researchers at Johns Hopkins say they’ve made a breakthrough that could help change that.

A 3D print of precancerous cells found in a human pancreas was created using CODA, the first-ever 3D genome profiling technique.

Doctors look at tissue in 3D and then map the genetic alterations onto the 3D model.

Never before have researchers been able to analyze precancers in this much detail. These precancers don’t show up on traditional radiology exams, which is why it’s difficult to detect pancreatic cancer early. Researchers say what they’re learning from this technology could eventually teach doctors how to spot the cancer sooner.

Here’s how it works: Researchers start with microscopic slides, each with a pancreas sample. They then scan all of those slides, so you have thousands of images.

Engineers use a combination of AI and coding to make visualizations to map the tissue.

Then, pathologists perform DNA sequencing to examine the cells and learn how some mutate into full-blown cancer.

“What are the features of these precancers that are more likely to progress, so then we know which ones to intervene on,” said Dr. Laura Wood, associate professor of pathology and oncology at Johns Hopkins Kimmel Cancer Center.

Another key finding is that most of us have these precancers in our pancreas, so researchers are exploring how it only becomes cancer in some of us.

The researchers are also now exploring whether this technique can help identify precancers in other organs as well.