Transpara

A study compared breast cancer detection performance between a panel of three unaided human readers (HR) and a stand-alone artificial intelligence (AI)-based Transpara system in Japanese women. The results showed that the AUC was higher for human readers (0.816) compared to the Transpara system (0.706), with human readers demonstrating a sensitivity of 89% and specificity of 86%. While the Transpara system’s performance was lower, with sensitivity of 93% and 85% at different cutoff scores and specificity of 45% and 67%, respectively, advancements in AI are anticipated to improve its performance and narrow the gap with human experts.