NewsLocal NewsKent


Artificial intelligence technology detects breast cancer early

Breast cancer AI technology
Posted at 11:44 PM, Jan 12, 2020
and last updated 2020-01-14 13:44:48-05

WYOMING, Mich. — Ground-breaking technology at one West Michigan hospital is detecting breast cancer for patients who may be in early stages of the illness.

Radiologists at Metro Health-University of Michigan Health have the impossible job of looking at thousands of scans everyday to detect cancer. When it comes to detecting breast cancer, radiologists look at about 400 images per patient. New technology is trying to isolate specific areas of concern for cancer risk.

"Some of these abnormalities are very hard to see," said Dr. Mark Triall, radiologist for Metro Health in Wyoming. "You might see it one day. You might not see it the next."

Dr. Triall has spent decades recognizing the smallest of details on mammograms.

"It's a lot of stress," said Dr. Triall. "This is something that you worry about reading these everyday, 'Did I see what I was supposed to read today, looking at literally thousands of images everyday?'"

In September, Dr. Triall's examined Gwennan Engen's scan, using new artificial intelligence technology.

After having a mammogram and other tests done, Engen learned she had stage 1 breast cancer.

"To say I was shocked is an understatement," said Engen.

In Engen's case, the AI technology circled an area that it identified as abnormal. It gave that area a 56% chance of being cancerous.

The AI technology is used in conjunction with a doctor to detect very early stage cancer that may be missed otherwise.

"It was nice to basically have that as a backup, and it basically proved that he is doing what he is supposed to do," said Engen.

"What makes it special for our field is that makes us better at what we do," Dr. Triall.

The AI technology only works for mammograms done with 3-D images. Not all insurance companies in Michigan cover 3-D images for mammograms. For the ones that do, this technology does not bring any additional cost.