A new artificial intelligence (AI) model could help physicians diagnose lung cancer earlier, according to a study published by Hunter et al in eBioMedicine. The findings suggested that the new model may yield a diagnosis more quickly and potentially more accurately than two existing risk assessment scores.
Background
Lung cancer is the leading cause of cancer mortality globally. Patients diagnosed with early-stage disease can be treated much more effectively, but recent data show that over 60% of lung cancer cases in England are diagnosed at either stage II or stage IV—highlighting the need for new initiatives to speed up detection.
Lung nodules are common abnormal growths that are mostly benign; however, some of them can be cancerous—with large nodules 15 mm to 30 mm in size often associated with the highest risk of becoming cancerous.
Study Methods and Results
In the new LIBRA study, researchers used data from the computed tomography (CT) scans of nearly 500 patients with large lung nodules to develop an AI model and test whether it was capable of accurately identifying cancerous nodules. To analyze the CT scan data, the researchers utilized radiomics to extract information about the patients’ nodules from medical images that can’t easily be detected by the human eye.
The researchers then conducted area under the curve (AUC) analyses to assess the AI model’s efficacy and discovered that it was able to identify each nodule’s risk of cancer with an AUC of 0.87. The performance was an improvement over the Brock score and the Herder score, which demonstrated AUCs of 0.67 and 0.83, respectively, for malignancy classification.
Further, since the new AI model uses only two variables—as opposed to nine for the Brock score and seven for the Herder score—it could potentially streamline and speed up nodule risk calculations in the future.
The new model may also help physicians make treatment decisions that currently don’t have clear referral pathways. Using the Herder score, patients are categorized as low risk if they score less than 10%, and high risk and requiring intervention if they score over 70%. For the patients in the intermediate risk group, scoring between 10% and 70%, a broad range of tests or treatment options could be considered. However, when combined with Herder score, the new AI model was capable of identifying high-risk patients in the intermediate group and suggested early invention for 82% (n = 18/22) of those who had large nodules that were then diagnosed as cancerous.
Conclusions
“According to these initial results, our model appears to identify [large] cancerous lung nodules accurately. In the future, we hope it will improve early detection and potentially make cancer treatment more successful by highlighting high-risk patients and fast-tracking them to earlier intervention,” emphasized lead study author Benjamin Hunter, MBChB, MRes, a PhD student in the Department of Surgery and Cancer at the Imperial College London and a clinical oncology registrar at The Royal Marsden National Health Service Foundation Trust. “Next, we plan to test the technology on patients with large lung nodules in [the] clinic to see if it can accurately predict their risk of lung cancer,” he noted.
“While at an early stage, this study is an example of the vital scientific clinical research we’re undertaking in the Early Diagnosis and Detection Centre at The Royal Marsden and [The Institute of Cancer Research]. Through this work, we hope to push boundaries to speed up the detection of the disease using innovative technologies such as AI,” underscored senior study author Richard Lee, MBBS, MRCP, PhD, Team Leader of the Early Diagnosis and Detection program in the Division of Genetics and Epidemiology at The Institute of Cancer Research and a consultant physician in the Department of Respiratory Medicine at The Royal Marsden National Health Service Foundation Trust.
“[Patients] diagnosed with lung cancer at the earliest stage are much more likely to survive for 5 years, when compared with those whose cancer is caught late. This means it is a priority we find ways to speed up the detection of the disease, and this study—which is the first to develop a radiomics model specifically focused on large lung nodules—could one day support clinicians in identifying high-risk patients,” Dr. Lee concluded.
Disclosure: The research in this study was supported by The Royal Marsden Cancer Charity, the National Institute for Health and Care Research, RM Partners, and Cancer Research UK. For full disclosures of the study authors, visit thelancet.com.
