CMET-04. RADIOMICS FEATURES CAN DIFFERENTIATE THE EGFR MUTATION STATUS OF BRAIN METASTASES FROM NON-SMALL CELL LUNG CANCER

Abstract

BACKGROUND AND PURPOSE: To assess whether radiomics features on DTI and conventional postcontrast T1-weighted (T1C) images can differentiate the epidermal growth factor receptor (EGFR) molecular status in brain metastases from non-small cell lung cancer (NSCLC). MATERIALS AND METHODS: Radiomics features (n = 5046) were extracted from preoperative MRI including T1C and DTI from pathologically confirmed brain metastases of 59 patients with underlying NSCLC and known EGFR mutation status (31 EGFR wild type, 28 EGFR mutant). A subset of 4317 features (85.6%) with high stability (intraclass correlation coefficient > 0.9) were selected for further analysis. After feature selection by the least absolute shrinkage and selection operator, the radiomics classifiers were constructed by various machine learning algorithms. The prediction performance of the classifier was validated by using leave-one-out cross-validation. Diagnostic performance was compared between multiparametric MRI radiomics models and single imaging radiomics models using the area under the curve (AUC) from ROC analysis. RESULT(S): Thirty-seven significant radiomics features (6 from ADC, 6 from fractional anisotropy [FA], 25 from T1C) were selected. The best performing multiparametric radiomics model (AUC 0.97, 95% CI 0.94-1) showed better performance than any single radiomics model using ADC (AUC 0.79, p = 0.007), FA (AUC 0.75, p = 0.001), or T1C (AUC 0.96, p = 0.678); the accuracy, sensitivity, and specificity of this model were 94.4%, 96.6%, and 92.0%, respectively. CONCLUSION(S): Radiomics classifiers integrating multiparametric MRI parameters may be useful to differentiate the EGFR mutation status in brain metastases from lung cancer.