miliary nodules
Credit: the rads/Getty Images

A ground-breaking randomized controlled study in Korea demonstrated that AI-based software designed to evaluate chest X-rays improved the detection of lung nodules. The results, published today in the journal Radiology, point a path to the use of this technology as regular part of clinical screening for lung nodules.

“Detecting lung nodules, a primary finding of lung cancer, is one of the crucial tasks in chest X-rays,” said Jin Mo Goo, MD, PhD, from the Department of Radiology at Seoul National University Hospital in Korea and a co-author of the paper. “Many studies have suggested that AI-based computer-aided detection software can improve radiologists’ performance, but it is not widely used.”

Lung nodules, abnormal growths in the lung tissue, are common and are usually the result of a prior infection. However, in some instances, they can be indicative of developing cancer, and X-ray images have been an important tool of physicians screening for the disease.

Increasingly, AI is being used to aid in the detection and diagnosis of a range of diseases and is well-suited for the evaluation a breadth of imaging technologies. One application in oncology is the use of AI to examine digital pathology images, which is rapidly gaining acceptance as a method that analyze tissue samples for the presence of disease. It is also a valuable when there is a high volume of samples that need to be evaluated to flag only those samples needing further evaluation by a pathologist.

For the most recent study on how AI evaluation of X-ray images could impact clinical practice in screening for lung nodules, the research team collected a cohort that included 10,476 patients with an average age of 59, who had undergone chest X-rays at a health screening center between June 2020 and December 2021. Patients completed a self-reported health questionnaire to identify baseline characteristics such as age, sex, smoking status and past history of lung cancer. Eleven percent of the patients were current or former smokers.

“As our trial was conducted with a pragmatic approach, almost all enrolled participants were included, which is a real clinical setting,” Goo noted.

Enrolled patients were randomly divided evenly into either the AI group, or non-AI group. The first group’s X-rays were analyzed by radiologists aided by AI while the second group’s X-rays were interpreted without the AI results. For the purposes of the study solid nodules larger than 8 millimeters in diameter and subsolid nodules with a solid portion of 6 millimeters in diameter were classified as actionable and would require a clinical follow-up according to lung cancer screening guidelines.

Analysis of the data showed that the detection rate of nodules using AI was nearly twice that of traditional methods (0.59% versus 0.25%). There were no differences in false referral rates between the two methods. In total, lung nodules were found in two percent of the patients.

The investigators noted that while older age and a history of lung cancer or tuberculosis were associated with positive reports, these and the other health characteristics did not have an impact on the efficacy of the AI system; this suggests the use of AI could be an effective screening aid across different populations, even for those with diseased or postoperative lungs.

“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,” Goo said.

Based on the positive results of their study of chest X-rays, the team next will turn its attention to form a similar study using chest computed tomography (CT) which will also identify clinical outcomes and efficiency of clinical workflow.

Also of Interest