[Lunit Press Release] On August 6, Lunit, a leading provider of AI-powered solutions for cancer diagnostics and therapeutics, announced that its AI-powered chest X-ray analysis software, Lunit INSIGHT CXR, has demonstrated superior performance in tuberculosis (TB) detection in a large-scale, independent study published in The Lancet Digital Health.
Despite being a curable disease, TB remains the world's second leading cause of death from a single infectious agent, with an estimated 10.6 million new cases in 2022. A major challenge in TB control is underdiagnosis, with 3.1 million cases going undetected that year. While chest X-rays are more sensitive than symptom screening, their effectiveness is often limited by variability in human interpretation and a shortage of radiologists in high-burden countries. This context underscores the critical importance of advanced AI assistance.
The study, conducted by Dr. Zhi Zhen Qin and a team of researchers from the Stop TB Partnership at UNOPS and Heidelberg University Hospital, evaluated 12 commercially available AI products for TB detection using data from South Africa's national TB prevalence survey. Lunit INSIGHT CXR achieved the highest area under the receiver operating characteristic curve (AUC) of 0.902, significantly outperforming other products.
Key findings for Lunit INSIGHT CXR include: ▲Highest AUC (0.902) among all tested products ▲89.9% sensitivity at 67.7% specificity, meeting WHO target product profile ▲89.5% sensitivity at 70.2% specificity, meeting WHO target product profile ▲Ability to maintain high sensitivity (>90%) across a wide range of thresholds.
The study, which included 774 participants (258 bacteriologically confirmed TB cases), is the first to comprehensively evaluate multiple AI products in a high TB setting.
"This landmark study not only showcases our AI's superior accuracy but also demonstrates its alignment with WHO standards," said Brandon Suh, CEO of Lunit. "Lunit INSIGHT CXR's ability to maintain high sensitivity across diverse populations and thresholds is particularly crucial in resource-limited settings, where every undetected case can have far-reaching consequences. By bridging the gap in TB diagnosis, we are contributing to a future where no case goes undetected, especially in regions where it matters most."
The study's findings have significant implications for TB screening in high-burden settings. Lunit INSIGHT CXR's robust performance, particularly its ability to maintain high sensitivity across various thresholds, allows for tailored screening strategies. This adaptability enables healthcare systems to optimize detection rates while managing available resources for confirmatory testing, potentially leading to more accessible and cost-effective TB programs in diverse healthcare settings.