Predictive value of imaging combined with clinical features for EGFR mutation in non-small cell lung cancer
Objective To investigate the predictive value of chest CT imaging combined with clinical fea-tures in epidermal growth factor receptor(EGFR)gene mutation in non-small cell lung cancer(NSCLC).Methods Preoperative clinical data and unenhanced thin-slice CT images of 210 patients with pathologically proven NSCLC were retrospectively analyzed.All NSCLC patients were divided into EGFR mutant group(140 cases)and EGFR wild group(70 cases),randomly assigned to training group and validation group(in a 8∶2 ratio),and tumor imaging features on CT plain scan images were extracted.Minimum absolute contraction and selection operator(LASSO)regression analysis was used to screen features,and an imaging omics prediction model was established.Clinical features(CT imaging features and patient clinical data)and imaging omics la-bels were included to construct a comprehensive prediction model.Based on the comprehensive prediction mod-el,a nomogram was drawn to realize model visualization and model validation.Receiver operating characteristic(ROC)curves,calibration curves,and decision curves(DCA)were plotted to evaluate the model's predictive performance and clinical utility.Results The AUC of the training group was 0.756;that of the verification group was 0.696;that of the clinical feature model was 0.811;and that of the verification group was 0.651.The integrated prediction model combined with the imaging label could improve the prediction efficiency of EG-FR mutation status,with the training set AUC=0.847 and the verification set AUC=0.740.Conclusion Compared with clinical features or imaging tags alone,the integrated model constructed by combining imaging tags and clinical features has better predictive efficacy in predicting NSCLC EGFR gene mutation,which is helpful to guide clinical treatment strategy.
万方数据
维普
