A recent study by researchers based in Singapore suggests that the use of artificial intelligence (AI) in imaging analysis has potential to detect lung cancer. The researchers reported their findings in the journal Lung Cancer.

Lung cancer management has seen recent advancements; however, late-stage diagnosis is common. Lung cancer is a leading cause of cancer-related deaths globally, but lung cancer screening to detect it early may be useful in limiting morbidity and mortality with this condition.

“Our review on the use of AI-based medical imaging for the early detection of lung cancer in lung cancer screening demonstrated a promising diagnostic performance,” the investigators wrote in their report.

Continue Reading

The use of AI in assessment of medical images has been an increasingly widespread practice, and past research has explored its use with lung cancer screening. The study investigators performed a systematic review and meta-analysis of existing literature on the use of AI in lung cancer to estimate the diagnostic test accuracy (DTA) of AI-based imaging in lung cancer screening.

The investigators examined 10 different databases for published and unpublished studies evaluating the performance of AI-based imaging methods in lung cancer screening. They also searched for articles that were identified in the reference lists of relevant articles. The Quality Assessment of Diagnostic Accuracy Studies-2 tool was employed to evaluate study quality, and the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach was used to evaluate the certainty of evidence for diagnostic tests. DTA was assessed based on sensitivity and specificity.

A total of 26 studies were included in the analysis, and these studies collectively included 150,721 images involving 66,502 patients. In their meta-analysis of these studies, the researchers calculated a pooled sensitivity of 94.6% (95% CI, 91.4%-96.7%) for AI-based imaging in lung cancer screening. The pooled specificity was 93.6% (95% CI, 88.5%-96.6%).

Evaluated subgroups showed similar levels of sensitivity and specificity; subgroups involved categories of region, data source, year of publication, and type of AI used. However, when the researchers evaluated the studies by GRADE criteria, they found that there was a very low certainty of evidence for sensitivity and specificity results in these studies overall.

The investigators provided recommendations to improve the robustness of AI models for lung cancer screening and their evaluation in studies. “With the shortfalls in mind, we urge future end-users, policy-makers, and commissioners to consider the incorporation of AI-based imaging into lung cancer screening programs to achieve high-quality evidence on DTA,” the investigators concluded in their report.


Thong LT, Chou HS, Chew HSJ, Lau Y. Diagnostic test accuracy of artificial intelligence-based imaging for lung cancer screening: a systematic review and meta-analysis. Lung Cancer. 2022;176:4-13. doi:10.1016/j.lungcan.2022.12.002