Lung nodules are common abnormal growths that typically form on the lungs due to previous lung infections but can rarely be a sign of lung cancer. Chest X-ray is a common screening method used to identifying lung nodules. Artificial intelligence (AI) can serve as a powerful tool to help identify lung nodules, particularly when radiologists have a high volume of cases. Now, a pioneering, randomized controlled study evaluating the effect of AI-based software in real clinical practice has found that AI significantly improved the detection of lung nodules on chest X-rays.

In order to identify the actual effect that AI has in clinical practice, researchers at Seoul National University Hospital (Seoul, Korea) conducted a study involving 10,476 patients with an average age of 59 years, who had undergone chest X-rays at a health screening center between June 2020 and December 2021. Patients were also asked to complete a self-reported health questionnaire for identifying baseline characteristics such as age, sex, smoking status and previous history of lung cancer. Within the group of patients, 11% were current or former smokers. The researchers randomly divided the patients evenly into two groups - AI or non-AI. Radiologists aided by AI analyzed the X-rays of the first group while the X-rays of the second group were interpreted without using AI.

Solid nodules with diameters either larger than 8 millimeters or subsolid nodules with a solid portion larger than six millimeters were identified as actionable, meaning that the nodule required follow-up based on lung cancer screening criteria. The researchers identified lung nodules in 2% of the patients. Their analysis showed that the detection rate for actionable lung nodules on chest X-rays was higher when aided by AI (0.59%) as compared to without AI assistance (0.25%). They found no differences in the false-referral rates between the AI and non-AI interpreted groups.

Older age and a history of lung cancer or tuberculosis were associated with positive reports, although these and other health characteristics did not impact the efficacy of the AI system. This indicates that AI can perform consistently across different populations, including those with diseased or postoperative lungs. The researchers now plan to conduct a similar study using chest CT which will also identify clinical outcomes and efficiency of workflow.

"Our study provided strong evidence that AI could really help in interpreting chest radiography. This will contribute to identifying chest diseases, especially lung cancer, more effectively at an earlier stage," said study co-author Jin Mo Goo, M.D., Ph.D., from the Department of Radiology at Seoul National University Hospital. 

http://www.gzjiayumed.com/en/index.asp