Skip to main content
. 2022 Jun 21;12:894696. doi: 10.3389/fonc.2022.894696

Figure 4.

Figure 4

The radiomics analysis procedures to build the classification model. Step1: The lesion ROI is outlined on each slice and then combined into a 3D tumor mask. Step 2: PyRadiomics is applied to extract 107 features, including shape, first-order statistics, and texture from each tumor mask (GLCM is an example of a feature processing). Step 3: The sequential feature selection is performed by using SVM, and finally, Step 4: The SVM algorithm is applied to build the classification model.