Fig. 8
From: Uncovering novel functions of NUF2 in glioblastoma and MRI-based expression prediction
Development and validation of a radiogenomics-based NUF2 expression model for GBM patients. (A,B) LASSO coefficient profiles of the radiomics features. The vertical dashed line indicates the optimal lambda value selected by 10-fold cross-validation. (C) Heatmap showing the association between NUF2 expression, clinical features (age, gender, MGMT status, 1p/19q codeletion), and overall survival status in GBM patients. (D) Variable importance plot showing the top radiomics features (RF1-RF6) ranked by MeanDecreaseGini in the random forest model. (E) Receiver operating characteristic (ROC) curve of the radiomics signature for predicting NUF2 expression in GBM patients. The area under the curve (AUC) was 0.897. (F) Calibration curve of the radiomics signature for predicting NUF2 expression in GBM patients. The dashed line represents the ideal calibration, and the solid lines represent the logistic and nonparametric calibration curves. (G) Decision curve analysis (DCA) of the radiomics signature compared to the “All” and “None” strategies. The net benefit was plotted against the threshold probability.
