Radiomics based on dual-layer spectral detector CT for predicting EGFR mutation status in non-small cell lung cancer.

Abstract

Objective: To explore the value of dual-layer spectral computed tomography (DLCT)-based radiomics for predicting epidermal growth factor receptor (EGFR) mutation status in patients with non-small cell lung cancer (NSCLC).

Methods: DLCT images and clinical information from 115 patients with NSCLC were collected retrospectively and randomly assigned to a training group (n = 81) and a validation group (n = 34). A radiomics model was constructed based on the DLCT radiomic features by least absolute shrinkage and selection operator (LASSO) dimensionality reduction. A clinical model based on clinical and CT features was established. A nomogram was built combining the radiomic scores (Radscores) and clinical factors. Receiver operating characteristic (ROC) analysis and decision curve analysis (DCA) were used for the efficacy and clinical value of the models assessment.

Results: A total of six radiomic features and two clinical features were screened for modeling. The AUCs of the radiomic model, clinical model, and nomogram were 0.909, 0.797, and 0.922, respectively, in the training group and 0.874, 0.691, and 0.881, respectively, in the validation group. The AUCs of the nomogram and the radiomics model were significantly higher than that of the clinical model, but no significant difference was found between them. DCA revealed that nomogram had the greatest clinical benefit at most threshold intervals.

Conclusion: Nomogram integrating clinical factors and pretreatment DLCT radiomic features can help evaluate the EGFR mutation status of patients with NSCLC in a noninvasive way.