AI can see if non-smokers are at high lung cancer risk  

Artificial intelligence could soon be used to scan lung X-ray images for signs of lung cancer in non-smokers, potentially identifying those at high risk. 

Lung cancer is the single biggest cancer-related killer worldwide and is caused by smoking in 85% of cases. That leaves 15% of the population potentially vulnerable to the disease, with risk factors including ‘second-hand smoke’, radon and asbestos exposure, air pollution and family history of lung cancer. 

Now, a new study presented at the annual meeting of the Radiological Society of North America has found machine learning models can be trained to spot signs of lung cancer on X-rays. 

X ray vs ct scan comparison of human lungs
Comparison between X-ray shows a small nodular opacity (arrow) in the left upper lung zone, detected with AI, compared to a non-contrast, low-dose chest CT scan showing a 9-mm solid nodule (arrow) in the left upper lobe. Credit: RSNA/Anika S. Walia

It could provide a simple screening measure for ‘never-smokers’, who could be vulnerable to the disease based on secondary risk factors.  

In the US, authorities recommend lung cancer screening for people aged 50-80 who have a smoking history of at least 20 packs a year. Australia is currently updating its processes for a July 2025 start date. 

The screening – usually by a low-dose CT scan – has a high rate of sensitivity and specificity in cancer detection but is not recommended for non-smokers. Boston University School of Medicine researcher Anika Walia believes AI could provide an effective solution to prevent these people from being excluded.  

She led the study which trained a deep learning model to scan lung X-rays for patterns associated with cancer. X-rays are considerably cheaper than CT. 

Named ‘CXR-Lung-Risk,’ the model was trained on nearly 150,000 chest X-rays of more than 40,000 asymptomatic smokers and never-smokers from a large US-based cancer screening trial to estimate lung cancer risk.  

“A major advantage to our approach is that it only requires a single chest-X-ray image, which is one of the most common tests in medicine and widely available in the electronic medical record,” says Walia. 

Of those included in the research, 28% were classified as being at high risk of lung cancer. 2.9% were later diagnosed with the disease. 

Dr Michael Lu, the director of AI at the Massachusetts General Hospital’s Cardiovascular Imaging Research Center, says the discovery provides an avenue to develop new screening methods as smoking rates continue to decline.  

“This AI tool opens the door for opportunistic screening for never-smokers at high risk of lung cancer, using existing chest X-rays in the electronic medical record,” says Lu.  

“Since cigarette smoking rates are declining, approaches to detect lung cancer early in those who do not smoke are going to be increasingly important.”    

Sign up to our weekly newsletter

Please login to favourite this article.