Fig. 4From: Lymph node metastasis prediction and biological pathway associations underlying DCE-MRI deep learning radiomics in invasive breast cancerAssociation of deep learning radiomics signatures. (A) Training cohort. (B) Validation cohort. Distribution difference of signatures in LN status of each cohort (Mann-Whitney U test, * P<0.05, ** P<0.01, ***P<0.001). Model signatures were correlated in the training cohort, especially in the radiomics and GoogLeNetnet signatures (PCC, rā=ā0.59, Pā<ā0.01), and also in the validation cohort (PCC, rā=ā0.32, Pā=ā0.01)Back to article page