Objective:To predict the Ki-67 expression grading in patients with mass breast cancer based on multimodal ultrasound features to aid clinical diagnosis and treatment.Methods:Ninety-three female patients (93 masses in total) with breast cancers confirmed by pathological examination were retrospectively included in the Second Affiliated Hospital of Harbin Medical University from September 2017 to September 2020. According to the immunohistochemical results, the patients were divided into Ki-67 high expression group (55 cases) and Ki-67 low expression group (38 cases). The qualitative and quantitative features from two-dimensional gray-scale ultrasound, color Doppler flow imaging (CDFI), shear wave elastography (SWE) and contrast-enhanced ultrasound (CEUS) images of all breast masses were retrospectively analyzed, differential features were analyzed based on logistic regression algorithm. ROC curves and Kappa test were used for the evaluation of diagnosis.Results:The univariate analysis revealed statistically significant differences between the two groups for conventional ultrasound features (size, shape, margins), SWE features (stiff rim sign, Eratio), and CEUS features (perfusion defect, IMAX) (all P<0.05). In the multiple logistic regression analysis, the margins, stiff rim sign, and perfusion defect were the independent factors for predicting the Ki-67 expression (all P<0.05). The performance of the predictive model was 0.882 (95%confidence interval of 0.798-0.940, P<0.05) with the sensitivity of 0.818 and specificity of 0.790. Conclusions:A preliminary analysis of the relationship between multi-modal ultrasound features and Ki-67 expression grading in mass breast cancers was performed based on logistic regression algorithm to provide more imaging information for clinical treatment and prognosis assessment.