Artificial Intelligence Support for
Mammography: In-Practice Clinical Experience
Authors Marie Tartar, MD, Lina Le, MD, Alyssa. T. Watanabe, MD, Alexander J. Enomoto, AA
Background: In June 2019, an outpatient imaging facility in Southern California implemented CureMetrix AI-based software for mammography in its cloud-based, Software-as-a-Service (SaaS) solution. CureMetrix AI-based CAD was run alongside traditional CAD, so radiologists could see the results in a real-time, head-to-head comparison.
Key Results and Conclusions
After two years following implementation, radiologists at this practice found that using CureMetrix AI delivered:
- A 71 percent reduction in flags over non-AI CAD, regardless of the patient’s breast density;
- Average turnaround times for callback reports that were nearly three times faster, expediting workups and giving patients quicker, more accurate results of their mammograms;
- Greater ease for radiologists in reviewing a filtered worklist preference (cmTriage) and an AI-CAD marked mammogram (cmAssist) over older, traditional CAD applications.
“After over two years of using CureMetrix AI, the physicians in this practice benefited from having fewer, but more meaningful flags to evaluate in a pre-sorted screening mammogram worklist,” said radiologist Dr. Marie Tartar, author of the study. “We are proud to publish these results in JACR, which we hope will inform more radiologists about the many possibilities of integrating AI CAD and mammography triage into their daily operations.”
CureMetrix cmTriage™ is the first FDA-cleared AI-based triage solution for mammography in the U.S. This workflow optimization tool enables a radiologist to sort and prioritize their mammogram worklist, flagging suspicious cases. Previous research has shown that cmTriage can help radiologists read mammograms up to 30 percent faster, streamlining their workflow.
CureMetrix cmAssist® is an AI-based diagnostic solution that helps radiologists identify, mark and score regions of interest in mammograms, finding potential breast cancers earlier. In retrospective clinical studies, cmAssist has demonstrated the ability to spot breast cancer up to six years before its initial first detection. Other studies have shown that cmAssist helps radiologists improve their cancer detection rate by an average of 27 percent, without an increase in recalls, and reduce false positives an average of 69 percent over traditional, non-AI CAD software.
PDF: Artificial Intelligence Support for Mammography: In-Practice Clinical Experience
Full Study: Artificial Intelligence Support for Mammography: In-Practice Clinical Experience