Table 2 Machine learning models for the diagnosis of axillary lymph node metastasis
References | Model | Imaging method | Sensitivity | Specificity | Accuracy | AUC |
|---|---|---|---|---|---|---|
Yang et al. [75] | radiomics model | CECT | 0.882 in the testing cohort 0.824 in the validation cohort | 0.887 in the testing cohort 0.963 in the validation cohort | 0.885 in the testing cohort 0.891 in the validation cohort | 0.94 in the testing cohort 0.92 in the validation cohort |
Liu et al. [93] | radiomics model | DCE-MRI | 0.901 in the training set 0.778 in the validation set | 0.833 in the training set 0.861 in the validation set | 0.896 in the training set 0.886 in the validation set | 0.914 in the training set 0.869 in the validation set |
Yu et al. [94] | clinical-radiomics nomogram | DCE-MRI | / | / | / | 0.92 in the development cohort 0.90 in the validation cohort |
Song et al. [95] | clinical-radiomics nomogram | DCE-MRI | 0.821 in the training cohort 0.759 in the validation cohort | 0.837 in the training cohort 0.845 in the validation cohort | / | 0.907 in the training cohort 0.867 in the validation cohort |
Ren et al. [97] | CNN model | MRI | 0.921 | 0.793 | 0.848 | 0.91 |
Chen et al. [98] | CNN model | DCE-MRI | 0.755 | 0.883 | 0.892 | 0.899 |