More Effective Mammography Through Artificial Intelligence and 3D Imaging
As women complete their annual mammogram, many in October as part of Breast Cancer Awareness month, those visiting TriHealth for the scan will have access to advanced technology.
“The detection of cancer has greatly improved in the last ten years, first with the introduction of 3D mammography, which improved the sensitivity for detection of cancer, and now with the introduction of artificial intelligence, or AI, technology,” said Anthony Antonoplos, MD, Radiologist with TriHealth. “In the past, during standard breast mammography, only two images of each breast, four images total, were used for review. 3D tomosynthesis allows the number of images captured during a standard mammogram to get as high as 250 to 300 images.” With this dramatic increase in the number of images collected, the complexity and challenge in interpreting results has also grown.
Profound AI technology, first approved by the FDA in 2018, is a new tool allowing for a deeper dive into the expansive results provided through 3D mammograms. “Each image captured by a 3D scan is important because a questionable finding may only appear on one or two of the images,” Dr. Antonoplos explained. “AI technology, in this case, works as a computer performing cognitive functions that a human would have done – it’s assisting the radiologist in reading and interpreting the images.” The computer works by placing a mark any place, on any image, for which there is a concern for a finding, assisting the individual reading a scan in quickly identifying where further attention may be needed. AI identifies suspicious results through a combination of tactics, including access to a library of previous 3D images where it can see patterns with a cancerous outcome, in addition to the technology’s cognitive ability. It learns as it reads and combines what it sees with previous patterns to formulate a red flag.
With all its benefits, the technology is not perfect on its own, the human factor of reading results is still immensely important. The combination of a radiologist working with AI technology has furthered increased the sensitivity of finding cancer by 8 percent. Not everything the AI marks as a questionable finding would require a call back. The technology assigns a score to anything flagged on an image, which reflects the strength of concern the AI has for what it sees, allowing radiologists to focus on the most important information with the most urgency. This score also assists in reducing false positive results, with a 7 percent reduction in call backs. A false positive occurs when a finding is identified as potentially developing cancer and a follow up diagnostic test does not result in a cancer finding. With the calls and discussions that accompany further testing is a heightened amount of stress and anxiety, which AI can contribute in reducing.
As Profound AI technology becomes more commonly used, patients will benefit further from the relationship between the computer and her or his radiologist. “As more and more scans are added to the AI access library, the technology will only become more effective in identifying and flagging concerning patterns,” Dr. Antonoplos said. “It will learn with its cognitive ability to not mark previous inconsequential findings, and it will have more cancerous patterns to study and understand, allowing us to find cancer sooner.”
TriHealth wants to remind everyone that October is Breast Cancer Awareness month. Remember for yourself and remind loved ones that screening saves lives. Schedule your mammogram today. Go to TriHealth.com/Mammo to learn more.
Last Updated: October 26, 2020