Lung nodules are tiny unwanted structures that can grow in human lungs, mostly due to previous lung infections. Although the nodules are found in the chest CT scans or X-rays of 50 percent of human adults, these abnormal growths also serve as important symptoms of lung cancer in some cases.
A team of researchers from Seoul National University (SNU), South Korea, claims that artificial intelligence can improve the chest X-ray-based lung nodule detection method.
To test the level of improvement that AI can bring in chest X-rays, the researchers conducted a study in real clinical settings involving 10,476 patients. Their experiment analyzed how AI software could help radiologists in the successful detection of lung nodules in large numbers, which is otherwise a difficult and time-consuming process.
How can AI improve lung nodule detection?
The researchers highlight that the detection of lung nodules in chest X-rays is important, especially for patients with lung cancer. Normally, lung nodules are 10 millimeters in size and don’t show much growth. However, if there is lung cancer, nodules grow more than 25 mm in size and show rapid progression.
According to a report from the American Thoracic Society, normal lung nodules in people who smoke have higher chances of turning into cancerous nodules than in those who don’t smoke. Early detection of cancer nodules is crucial for the effective treatment of lung cancer patients.
When it comes to affecting the human population, lung cancer ranks second (after breast cancer) among all cancer types. It killed 1.79 million people globally in 2020 alone. The early detection of cancerous lung nodules increases the chances of a patient surviving this deadly disease, and many scientists across the globe believe that AI can improve lung cancer screening.
While explaining the benefits of AI, the co-director of education at the University of California San Francisco Center for Intelligent Imaging, Jae Ho Sohn (not involved in the current study) said in a 2021 news release:
“AI algorithms will miss nodules that humans will detect, but it also detects nodules that humans may miss. (Plus,) when an AI algorithm detects nodules and puts a caliper on the size of the nodules, it’s less prone to precision errors from the ruler being offset a little bit. Also, algorithms will tend to be much more consistent each time compared to a group of human radiologists. This results in more accurate nodule size reporting.”
The researchers at SNU argue that although many previous studies have highlighted the potential of AI in X-ray nodule detection, it is still not practiced widely because of a lack of evidence. Their experiment aims at providing solid evidence on whether or not AI is useful for the purpose.
Using AI to screen lung cancer for real
The study authors examined 10,476 patients and asked them to fill out a questionnaire to collect information related to their age, gender, past lung infection history, and smoking behavior. 11 percent of the total subjects were either smokers or used to smoke earlier in their lives. The patients were divided into two groups and their chest X-rays were performed.
The X-rays of the first group were studied by radiologists who were assisted by AI software, and the X-rays of the second group were examined only by radiologists. Lung nodules of more than 6 mm in size were detected in about two percent of the total subjects.
Interestingly, nodule detection was higher in the first group (0.59 percent) that employed AI, as compared to the second group (0.29 percent). These results suggest that AI improves our chances of detecting nodules in the lungs. While commenting on these results, one of the study authors and a professor at SNU Medicine, Jin Mo Goo said:
“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”
Professor Goo and his team plan to conduct more such studies in the future, to further highlight the outcomes of using AI-aided lung nodule detection in chest X-rays.
The study is published in the journal Radiology.