AI Software Analyzes Mammograms with High Accuracy

A radiology artificial intelligence (AI) algorithm provides the location of lesions suspicious of breast cancer and an abnormality score that reflects the AI's confidence level.

The Lunit (Seoul, South Korea) Insight MMG algorithm is intended to aid detection, localization, and characterization of suspicious areas for breast cancer on 4-view Full-Field Digital Mammography (FFDM) images. The location of breast cancer is generated in the form of heatmaps and/or contour maps. The AI abnormality score, derived without human interpretation, reflects the probability of the existence of breast cancer, and also provides an assessment of breast density.


Major benefits include improved detection accuracy, reducing false-positive recalls by 14% and false-negative recalls by 18%; Automatic triage of up to 60% of the cases based on AI abnormality scores; improved reading performance by general radiologists, to a level of breast specialists; early diagnosis of T1 and node-negative breast cancer with 91% and 87% accuracy, respectively; decision-making support in complicated, hard-to-conclude cases classified as BI-RADS 3 or 4; and improved diagnostic accuracy for dense and fatty breasts by up to 9% and 22%, respectively.

“With our AI solution, we hope to increase the efficiency and accuracy of mammography screening. We can assist radiologists diagnose diseases at an earlier stage, helping patients be treated at the right time,” said Brandon Suh, CEO of Lunit. “We strongly believe that sooner or later AI will become the new standard of care, and AI will be used everywhere as a must-use product. Not only will it be used as a tool to make the workflow more efficient, but it will ensure better diagnosis and healthier life for patients.”

Related Links:
Lunit