Development of a Tumor and Peritumor Radiomics-based Model to Predict the High-risk Surgical Margin and Prognosis of T3 and T4 Non-Small Cell Lung Cancer

The investigators retrospectively collected the participants with T3 and T4, N0-2, M0 NSCLC patients resected between January 2013 to December 2021 for training and internal validation. The Clinical data, preoperative laboratory results and images were collected. High-risk margins were defined as R1 or R2 surgical margins or local recurrence during follow-up, and the investigators also collected the disease-free survival time. On the Deepwise multi-modal research platform, the images were semi-automatically segmented and expanded outward by 3mm to obtain the peritumor tissue. PyRadiomics was used to extract the radiomic features. LASSO was used to select the features and tumor radiomics model, peritumor model and combined model were built using 5-fold cross-validation. And it was further tested on the independent cohort. Discrimination was assessed by using the C-index and area under the receiver operating characteristic curve (AUC), sensibility, specificity.