Lung cancer screening—Marching along …

capability through analysis of an entire 3D computed tomography (CT) data set, rather than just a sequence of 2D slices, interrogating the tumour environment, such as blood vessels and other lung structures, and radiomics, the quanti-tative, rather than visual, analysis of nodule features. The data sets required to train these models are very large, requiring many thousands of annotated CT scans. To this end, large data sets are now publicly available. Screening programmes may be well placed to prospectively curate imaging data sets, or at least contribute to international repositories. This is important because lung cancer is a mov-ing target; we have already witnessed the evolution from predominant central squamous cell carcinomas to more peripheral adenocarcinomas in response to changes in ciga-rette manufacture and design over the past half-century. Constant product innovation from tobacco not least of which, the increasing use of electronic cigarettes by younger cohorts of people, will no doubt lead to future chal-lenges for lung cancer.