Lung Nodule Characterization by Artificial Intelligence Techniques

Management of incidental lung nodule is difficult and mainly based on simple morphometric characteristics such as maximum size and shape. Radiomics could play a role in simplifying this management by orientating towards a benign or a malignant origin, by comparing advanced characteristics to a large database of lung nodules. The primary purpose is to evaluate the performances of a novel tool based on radiomics to characterize incidental lung nodules, discovered on computed tomography. The secondary objectives are: * to evaluate the variation in the performances of the software based on various technical aspects of the CT, such as radiation dose, reconstruction algorithm, type of scanner,… * to compare the performances of this software to those of expert readers, * to analyze the potential impact of this software on patient's management.