AI

computer data stream

A Utah Valley University professor hopes his research will help catch cancer years before patients begin showing symptoms.

“The holy grail of cancer treatment going forward isn’t the problem of treating the cancer, but how do we find the cancer sooner,” said Vern Hart, an assistant professor of physics at UVU.

For the last four years, Hart and a group of about 60 students have been shining infrared lasers through both healthy and cancerous cells in order to create a database for an artificial intelligence program to learn how to identify different types of cancerous cells.

Cancerous cells and healthy cells diffract light differently, and those differences may allow cancer to be caught when there’s still trace amounts of it in the bloodstream.

Hart, who is pursuing publication of his research, previously worked in the radiation department at a Milwaukee cancer center. He said most cancer cases aren’t diagnosed until someone complains of symptoms, and there isn’t a common technique to find cancer before there’s a tumor.

“Detecting cancer at the cellular level isn’t a common practice right now,” Hart said.

His team used the AI neural network to find patterns in the images of seven different types of cancer, and found it was more than 90% accurate at identifying cancer cells during its first tests.

Hart said he was shocked at how accurate it was so early in the research process.

“It usually doesn’t work the first time, or the fifth time, or the sixth time,” he said.

Cancer cells look different from healthy cells. Hart said the nucleus is typically larger and elongated, which changes how the cell scatters light when hit by a laser. The AI program reads the light pattern that emerges on the other side of the cell and can identify which type of cancer cell it is.

“We’ve had some really good success with any two cells that are very different from each other in shape,” Hart said.

What comes next is to test the system with mixes where only 10% of the cells in a sample are cancerous and the rest are healthy. Hart hopes his research will lead to noninvasive ways to detect cancer cells that can cut down on procedures like colonoscopies and biopsies.

Right now, he considers locating trace amounts of cancer in the blood to finding a needle in a haystack. But, he said, the technology could raise warning flags that someone needs to be checked for cancer.

“You can erase a lot of the invasive procedures, just make it a lot less invasive, and a lot more convenient for the patient,” Hart said.

See what people are talking about at The Community Table!