A research review looked at whether computer analysis of regular CT scans could predict hidden cancer spread in early-stage lung adenocarcinoma. The analysis combined data from 10 studies involving 6,349 patients. Researchers used a technique called radiomics, which extracts detailed measurements from medical images that the human eye can't see.
The analysis found that when tested on data from the same hospitals where the method was developed, the computer analysis correctly identified about 85% of patients with hidden spread and correctly ruled it out in about 78% of patients without it. When tested on data from different hospitals, the accuracy was lower—correctly identifying about 72% with spread and ruling it out in about 75% without it.
The main reason for caution is that the performance dropped when tested on different patient groups, suggesting the method might not work as well everywhere. The accuracy also seemed to vary depending on how the CT scans were taken and how the images were analyzed. This means the approach is still in the research phase and needs more standardized testing before doctors could reliably use it to decide whether patients need additional lymph node surgery.