New technology shows potential for identifying lung abnormalities at Royal Brompton Hospital

A study led by Royal Brompton Hospital has evaluated automated imaging technology that could help identify lung abnormalities in patients undergoing screening for lung cancer. The findings highlight how the technology could support clinicians to detect these abnormalities more consistently.

Interstitial lung disease (ILD) is a term used to describe a group of conditions that can cause inflammation and scarring of the lungs, which can affect how well the lungs work.

Interstitial lung abnormalities (ILAs) are subtle changes in the lung tissue that can be seen on CT scans. CT scans use X-rays to create detailed images of the inside of the body. ILAs can be a sign of inflammation, scarring or other changes in the lungs. The presence of ILAs is associated with an accelerated decline in lung function and an increased risk of mortality.

More than half of people with ILAs show progression of these changes on subsequent scans, with some developing clinically relevant pulmonary fibrosis, making their identification during lung screening potentially important.

Identifying ILAs can be challenging as assessments can vary between clinicians, highlighting the need for more standardised ways of identifying and assessing these abnormalities. An automated approach could help make this process more consistent.

A commercially available tool called e-Lung by Brainomix can be used to measure patterns associated with ILD on CT scans of the chest. The technology has previously been validated for assessing ILD, but its potential for identifying ILAs has not previously been established.

The study used this tool to assess CT scans from a lung cancer screening programme and explore whether the technology could accurately identify ILAs.

For the study, CT scans that had been reported by one of five consultant thoracic radiologists, each with a minimum of 8 years’ experience, were included. The radiologists had identified whether an ILA was present and, where present, estimated how much of the lung was affected.  

The e-Lung technology analysed the same scans and produced measurements of different lung patterns and the overall extent of abnormalities. The researchers aimed to determine the optimal thresholds for these metrics for identifying ILAs. They then compared the e-Lung results with the radiologists’ original reports to assess how accurately the technology identified ILAs.

A total of 171 scans were evaluated, including 70 in which ILAs had been identified and 101 normal scans without ILAs. The automated measures performed well in distinguishing scans where ILAs affected at least 5% or more than 10% of the lung from scans without ILAs.

The study shows that e-Lung has potential to support the identification of ILAs in people undergoing lung cancer screening. Further research in larger groups of patients is now needed to validate the findings and understand the impact of using the technology in clinical practice.

The study has been published in BMJ Open Respiratory Research and co-authored by several researchers at the hospital, including consultant radiologists Professor Anand Devaraj, Dr Emily Bartlett and Professor Sujal Desai, and respiratory consultants Dr Richard Hewitt and Professor Peter George.

Commenting on the research, Professor George said:

“Picking up the earliest changes on CT scans before symptoms emerge is likely to be crucial as we focus on improved outcomes for patients with interstitial lung disease. Lung cancer screening programmes represent an ideal setting for us to opportunistically identify these individuals.

“Artificial intelligence can ensure that these limited abnormalities on CT are quantified and then tracked over time so that diagnoses are reached and treatments are initiated based on objective and robust data. I am pleased that the Brompton is pioneering the use of this cutting-edge technology, maintaining its role as a global leader in the field.”