PACS, AI combination helps uncover low bone mineral density

The University of Virginia Health System (UVA) is currently testing a new AI platform with its current PACS program to assess patients for abnormal bone density as well as fatty liver, emphysema and coronary artery disease.

UVA currently uses VuePACS from Carestream and has added Zebra Medical Systems’ AI platform, Radiology Assistant, to the integrated system.

Gaskin noted that UVA has found the pairing of technologies is helping radiologists identify cases they could not detect visually, with special emphasis on bone mineral density at its earliest stages.

“In our young project, we have already seen a number of body CT reports suggesting abnormally low bone mineral density due to algorithm results,” said Christopher Gaskin, MD professor of radiology, medical imaging and orthopedic surgery, vice-chair of operations and informatics, and associate chief medical information officer at the University of Virginia Health System. “This should lead to formal bone density exams to confirm the suspected diagnosis of osteoporosis, that may lead to earlier treatment which helps reduce the risk of future osteoporotic fractures.”

Though radiologists typically would not miss the aforementioned disorders, Gaskin believes the algorithms are an additional check that makes it less likely for findings to be missed. The only exception, he says, is bone mineral density. The algorithm results could make it more likely that osteoporosis is detected because radiologists traditionally are not focused on bone mineral density.

“The subtler the finding is, the greater the likelihood of failed observation,” Gaskin said.

The system indicates a normal image with a green icon and abnormal results are indicated with a red icon. This approach has the potential to facilitate reviews of a large numbers of images, while also reducing the error rate in reports.

“It’s exciting to think that in the near future machine algorithms might help us in our work, reducing some of our burdens so we can focus on cognitive tasks requiring our expertise,” Gaskin said. “We will participate in the development and deployment of these technologies, and enjoy the benefits in the years ahead.”