This system utilizes a hybrid deep learning model, developed using high-quality clinical data from the lung cancer early screening program at National Taiwan University Hospital. It can accurately identify the diverse and minute nodule types required by Lung-RADS.
Rigorously validated by the TFDA, the AI's efficacy achieves a high sensitivity of 92% at a top-tier performance level—with only 0.6 false positives per scan, even on minute lung nodules as small as 4mm. It also supports AI-powered comparative tracking and structured reporting, and can be seamlessly integrated into hospital PACS/RIS.
LibraLung was developed to meet the immense demand of Taiwan's national lung cancer screening program. Jointly created by Cohesion Data and National Taiwan University Hospital, it has received TFDA medical device clearance and is the exclusive AI solution recommended by the Health Promotion Administration (HPA) on its official website.
Clinically proven to significantly enhance early diagnosis (a 126.7% increase in the interception rate of Lung-RADS 4 cases) and nearly triple physician workflow efficiency.