In a prospective clinical study, Lunit’s AI solution for chest X-rays more than doubled the rate at which lung nodules were detected in patients during routine health checkups.

Lunit’s INSIGHT CXR traces the location of lesions with an abnormality score that indicates the AI technology’s confidence. It can detect 10 of the most common chest abnormalities, including tuberculosis, with 97%–99% accuracy, says the company.

After showing promise in past retrospective studies, the new findings illustrate the AI-based CAD software's accuracy in real-world populations.

"This is the first real-world evidence proving that AI for chest radiography can improve actionable nodule detection without increasing false positives. We believe that this prospective study will lay the groundwork for AI to eventually become the standard of care for chest radiography,” said Lunit CEO Brandon Suh in a statement.

Out of 10,476 adults, South Korean researchers randomly assigned half to have their chest X-ray screenings assessed by the technology and radiologists, while the other half had theirs assessed without AI.

Lung nodules were identified in 2% of cases, with those by AI at a rate of 0.59%, versus those without at 0.25%. Detection was also higher in the AI group for malignant lung nodules, at 0.15% versus 0% in the non-AI group.

The trial for Lunit INSIGHT CXR was a pragmatic, randomized control one, with almost all enrolled patients included, making it a real clinical setting.

No significant difference was found in false-referral rates between both groups, and old age and a history of lung cancer or tuberculosis did not impact the efficacy of Lunit’s AI solution, indicating that it is consistent across different populations.

“This will contribute to identifying chest diseases, especially lung cancer, more effectively at an earlier stage,” said study co-author Dr. Jin Mo Goo, from the department of radiology at Seoul National University Hospital.

The solution has the CE mark and debuted in 2019 at the Radiological Society of North America Annual Meeting.

The findings were published in Radiology.

X-Ray Homepage