Fig. 2

The workflow of Radiomics analysis procedures. Step 1 Segmentation from cross-image: A represent the outline of MG, a 34-year-old male diagnosed A2B1L1 according to Montreal classification (tCDEIS = 4.5); B presents the outline of UG, a 65-year-old woman diagnosed A3B2L3 according to Montreal classification (tCDEIS = 7.5). Step 2 Feature selection and screening: Feature types include shape features, GLCM, GLDM, GLRLM, GLSZM, NGTDM, and mutual information was selected for dimensionality reduction. Step 3 The modeling approach: Logistic analysis was selected to model the retained features. Step 4 Model assessment: ROC curve, calibration curve and clinical decision curve were selected for assessing the efficiency of the radiomics model. (Note: GLCM = Gray Level Co-occurrence Matrix, GLRLM = Gray Level Run Length Matrix, GLSZM = Gray Level SizeZone Matrix, GLDM = Gray Level Dependence Matrix, NGTDM = Neighborhood Gray-Tone Difference Matrix)