Skip to main content
. 2019 Jun 14;9:505. doi: 10.3389/fonc.2019.00505

Figure 3.

Figure 3

The flowchart for feature engineering of multiparametric MR radiomics. (A) The whole-tumor segmentations from a total of 10 sequence images were executed. (B) A total of 2,498 features were extracted, and a two-stage feature selection method was subsequently performed. (C) Machine learning-based classifiers were used for the 4-IHC classification and TN vs. non-TN cancers.