SEOUL, South Korea, Nov. 25, 2024 /PRNewswire/ -- Lunit (KRX:328130.KQ), a leading provider of AI-powered solutions for cancer diagnostics and therapeutics, today announced its participation in the Radiological Society of North America (RSNA) 2024 Annual Meeting, presenting a record-breaking 20 abstracts. As RSNA 2024 highlights the transformative role of AI in radiology, Lunit's groundbreaking research and solutions emphasize how AI can address critical challenges in healthcare, from improving diagnostic accuracy to enhancing clinician efficiency.

Among Lunit's studies, two groundbreaking research presentations stand out:

1. Enhancing Breast Cancer Detection Across Diverse Populations

Presented by Dr. Hari Trivedi from Emory University, the study focuses on Lunit INSIGHT DBT, an AI-powered solution for breast cancer detection using digital breast tomosynthesis (DBT) images. Conducted on a large, racially heterogeneous screening population from Emory University (137,460 cases from 2013-2020), the research demonstrated Lunit AI's consistent performance across various subgroups, including race, ethnicity, age, and breast density.

Key findings include:

Robust overall AUC of 0.920
High sensitivity (84.5%) and specificity (83.8%)
Consistent performance across lesion types, from calcifications to architectural distortions
Demonstrated the reliability and potential of Lunit INSIGHT DBT to address disparities in breast cancer detection, underscoring its capability to serve as a robust tool for diverse patient populations globally

2. Empowering Clinicians Through AI-Assisted Chest X-Ray Interpretation

Presented by Dr. Ruchir Shah from Oxford University Hospitals and awarded the Trainee Research Prize - Resident, this study evaluated the impact of Lunit INSIGHT CXR, an AI-powered solution for chest X-ray interpretation, on clinician performance in emergency and inpatient care settings.

The study involved 30 clinicians from various specialties and experience levels, who interpreted 500 chest X-rays with and without AI assistance on the RAIQC platform.

Notable results include:

AI demonstrated superior standalone performance with AUCs of 0.83-0.99 across 10 pathologies, with exceptional accuracy (AUC>0.9) in 8 pathologies
Significant improvement in clinicians' accuracy in 8 out of 10 pathologies with AI assistance, including pulmonary nodules, pleural effusion, and fibrosis
Marked greatest improvement in fibrosis detection with a delta AUC of 0.193

Medical Research Homepage