Description

Early detection and diagnosis of pulmonary nodules is clinically significant regarding optimal treatment selection and avoidance of unnecessary surgical procedures. Deferential pathology results causes widely different prognosis after standard surgery among pulmonary precancerous lesion, atypical adenomatous hyperplasia (AAH) as well as adenocarcinoma in situ (AIS), and early stage invasive adenocarcinoma (IAC). The micro-invasion of pulmonary perifocal interstitium is difficult to identify from AIS unless pathology immunohistochemical study was implemented after operationwhich may causes prolonged procedure time and inappropriate surgical decision-making. Key feature-derived variables screened from CT scans via statistics and machine learning algorithms, could form a radiomics signature for disease diagnosis, tumor staging, therapy response adn patient prognosis. The purpose of this study was to investigate whether the combined radiomic signature based on the focal and perifocal5mmradiomic features can effectively improve predictive performance of distinguishing precancerous lesions from early stage lung adenocarcinoma. Besides, immunotherapy response is various among patients and no more than 20% of patients could benefit from it. None reliable biomarker has been found yet expect Programmed death-ligand 1 (PD-L1) expression, the only approved biomarker for immunotherapy. However recent reports suggested that patients could benefit from immunotherapy regardless of PD-L1 positive or negative. On the contrast, radiomics has show it advantages of non-invasiveness, easy-acquired and no limitation of sampling. Therefore, we applied this strategy in prediction for the immunotherapy response of advanced NSCLC lung cancer patients receiving immune checkpoint inhibitors (ICIs), which would prevent some non-benefit patient from the adverse effect of ICIs.