The diagnostic value of dynamic contrast-enhanced MRI radiomics combined with radiological features for the patho-logical grading of invasive breast cancer

CHEN Ruiyu; HUANGFU Fenghua; QIAN Weina; CAI Jie; LI Jian

doi:10.20258/j.cnki.1006-9011.2025.10.020

Abstract：Objective To investigate the diagnostic value of dynamic contrast-enhanced MRI(DCE-MRI)and radiological features in the pathological grading of invasive breast cancer.Methods 180 cases of invasive breast cancer were selected,in-cluding 38 cases of grade Ⅰ,47 cases of grade Ⅱ and 95 cases of grade Ⅲ.DCE-MRI was performed on the patients,and the im-aging characteristics of patients with different pathological grades were compared.Logistic regression analysis was used to screen the imaging characteristics related to pathological grades,and an imaging feature model was established to predict the pathologi-cal grades of invasive breast cancer.The AK software was used to extract the iconographic features from the DCE-MRI images,and the impact of the model on the histological characteristics of invasive breast cancer was established to predict the pathological grading.The ROC curve was used to analyze the predictive value of the imaging histological model,imaging characteristic model and their combination on the pathological grading of invasive breast cancer.Results The maximum diameter of tumor,unclear tumor boundary,burr sign and irregular shape were independent predictors of higher pathological grade of invasive breast cancer(P<0.001).The imaging characteristics model was established based on the above imaging characteristics,and its AUC for pre-dicting pathological grade of invasive breast cancer was 0.846.A total of 6 optimal imaging characteristics were screened from DCE-MRI images,and an imaging characteristics model was constructed.The AUC of this model for predicting the pathological grading of invasive breast cancer was 0.860.The AUC for predicting the pathological grade of invasive breast cancer by the combi-nation of imaging characteristics and imaging characteristics was 0.935,which was higher than that predicted alone.Conclu-sion The iconographic features acquired by DCE-MRI have a high predictive effect on the pathological grading of invasive breast cancer,and the combination with the imaging features can significantly improve the clinical diagnostic ability for the pathological grading of invasive breast cancer.