Distinguish EGFR+ and EGFR- patients in LA using CT images

Many reports show that lung adenocarcinoma (LA) is currently diagnosed at the advanced stages with a lower survival rate, and highly sensitive to the epidermal growth factor receptor (EGFR) gene mutation status. Therefore, great research has been made to implement lung cancer screening programs using computed tomography (CT) imaging modality for early detection of disease. This study aims to distinguish EGFR+ and EGFR- patients in LA using 2D and 3D CT image features in conjunction with forward feature selection and SVM. Focusing on the case of the LA patients data with EGFR mutation, experiment results show that the proposed approach can yield effectively discriminatory power to distinguish the EGFR mutation subtypes. Investigating other reproducibility of quantitative CT imaging features, such as pixel histogram, co-occurrence, Law's Masks and wavelet feature categories, as well as collecting more patients data are interesting future work.