AIM: To investigate the utility of contrast-enhanced mammography (CEM) in the differential diagnosis of breast non-mass enhancement (NME) lesions, and to compare the diagnostic performance of BI-RADS for NME lesions in MG, US, MRI, and CEM. MATERIALS AND METHODS: A total of 120 patients were included from April 2019 to April 2022. The independent enhancement descriptors for malignancy of NME were identified through binary logistic regression analysis. The modified BI-RADS in CEM (mBI-RADS (CEM)) was defined after adjusting the BI-RADS (MG) by independent descriptors. The diagnostic performances of BI-RADS in low-energy image, US, MRI, and CEM were evaluated and compared using receiver operating characteristic (ROC) analysis and DeLong's test. RESULTS: The enhancement descriptors of enhancement morphological distribution, internal enhancement pattern and time-signal intensity curve were identified as independent risk factors for malignancy (all P<0.001). For all-group NME analyses, the areas under the curve (AUCs) of mBI-RADS (CEM) were higher than those of BI-RADS (MG) (all P<0.050). For subgroup NME analyses, the AUCs of mBI-RADS (CEM) and BI-RADS (MRI) were higher than those of BI-RADS (MG) and BI-RADS (US) (all P<0.001). While in NME with nonemicrocalcification sub-group, the area under the receiver operating characteristics curve of mBIRADS (CEM) was higher than those of BI-RADS (MRI), both of which were higher than those of BI-RADS (MG) and BI-RADS (US) (all P<0.050). CONCLUSION: The mBI-RADS (CEM), which integrates morphological and enhancement features, can significantly improve the diagnostic accuracy compared with BI-RADS (MG). Furthermore, the mBI-RADS (CEM) demonstrated specific advantages for NME with microcalcifications compared with BI-RADS (MRI). (c) 2025 The Royal College of Radiologists. Published by Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.