Breast AI Research Portfolio

Whether you are reading DBT studies in a single reader screening environment like the United States, or identifying lower-risk studies for reducing reading burden in a double reading screening environment, Transpara saves practicing radiologists time – without compromising on clinical quality.

Reading faster has no value without confidence in the results. Transpara’s unmarked studies (70% of a typical screening population) have a 99.97% negative predictive value. These results save ~26% of reading time on DBT studies (Van Winkel, et al, 2021) or up to 44% of workload in a FFDM double-reading environment (Hernström, et al, 2025) while improving the cancer detection rate. Time saved can also save lives.

One of the barriers to confident mammography reading is breast density. Human sensitivity for breast cancer on mammograms can drop from around 80% in the least dense breast tissue to between 50-60% in the densest breasts. Transpara analysis is not affected by the visual characteristics of density in the same way.