Specific Aim 1. Use segmentation and quantitative analysis of lung lobes, airways and vessels to measure lung parenchyma features in addition to lung nodule features (in those with lung nodules) from high risk smokers who did or did not develop lung cancer within 5 years of the baseline LDCT to identify imaging features and patterns associated with lung cancer and no lung cancer.

Specific Aim 2. Determine the incremental value of lung parenchymal features compares to lung nodule features to determine malignancy risk within 5 years of baseline LDCT. We appreciate that in the NLST the nodules were not individually followed prospectively. However, in the Pan-Canadian Early Detection of Lung Cancer Study, which did follow individual nodules prospectively, we found that the largest nodule accounted for lung cancer in 80% of cases and the second largest nodule accounted for the lung cancer in 18% of cases. In addition, Dr. Tammemagi in malignancy risk modeling of NLST individuals with abnormal suspicious screens found that the largest nodule was most useful in modeling risk, much more useful than the second largest nodule. We will use these observations to inform our planned analysis.

John Mayo MD, FRCPC University of British Columbia
Martin Tammemagi PhD Brock University
Bram van Ginneken MSc. Radboud University Medical Center