Prediction of EGFR mutation status in non-small cell lung cancer based on multiparametric MRI radiomics

The purpose of the study was to establish and validate a model for predicting the mutation status of epidermal growth factor receptor (EGFR) in non-small cell lung cancer (NSCLC) using magnetic resonance imaging (MRI) radiomics features combined with clinicopathological factors. Overall, 91 patients with NSCLC (72 in the training cohort and 19 in the validation cohort) were included in this study; 1708 radiomics features were extracted from the MRI (T2W and CET1w) sequences. The variance threshold method combined with the univariate selection method and the least absolute shrinkage and selection operator (LASSO) regression was used to screen important radiomics features, calculate radiomics scores, and construct a radiomics model. Multivariate logistic regression analysis was used to combine radiomics scores (Rad-scores) and independent predictive factors to construct a radiomics nomogram for predicting EGFR mutation status. The predictive performance and clinical practicality of the model were evaluated using the area under the curve (AUC), calibration curves, and clinical decision curves. EGFR mutations were identified in 30.8% (28/91) of patients; 854 radiomics features were extracted from T2WI and CET1w, making a total of 1708 features. First, the variance threshold method was used to screen out features with variance < 0.8, yielding 1702 features. Features with an insignificant differences (p ≥ 0.05) were screened out using the univariate selection method, yielding 43 features. Finally, all features were fitted based on the type of gene mutation using the LASSO algorithm. Thirteen important radiomics features were screened. The radiomics model based on T2WI combined with CET1w provided better classification of EGFR mutant and wild-type, with AUCs of 0.846 and 0.808 in the training and validation cohorts, respectively. The radiomics nomogram model based on T2WI–CET1w radiomics label combination and independent predictors (gender and maximum diameter) for multivariate logistic regression analysis showed higher diagnostic efficiency, with AUCs of 0.880 and 0.859, respectively. The calibration curve showed good predictive performance, and the decision curve indicated that the radiomics nomogram had high clinical benefits. The model based on MRI radiomics shows strong diagnostic efficacy for predicting EGFR mutation status in NSCLC, guiding individualized targeted therapies.