The Value Of Multimodal Magnetic Resonance Imaging For Benign And Malignant Breast Cancer Discrimination

Objective To investigate the diagnostic value of predictive models based on dynamic contrast Enhanced MRI（DCE-MRI）,synthetic MRI（sy MRI）,and multiplexed sensitivity encoding（MUSE）diffusion-weighted imaging（DWI）sequences in the differentiation of benign and malignant breast lesions.Methods Clinical and MRI data of 161 patients with 166 breast lesions who are postoperative pathologically confirmed between September 2020 and December 2022 were retrospectively collected.All patients underwent DCE-MRI,MUSE-DWI,and sy MRI sequences before surgery.The lesions were classified according to BI-RADS.The apparent diffusion coefficient（ADC）values of MUSE and quantitative relaxation parameters[longitudinal and transverse relaxation times（T1,T2）,and proton density（PD）values]of sy MRI were measured,and the parameter variation values(T1_Delta,T2_Delta,PD_Delta)and change in their ratios(T1_ratio,T2_ratio,PD_ratio)were calculated.The lesions were randomly divided into training（n=117）and validation（n=49）groups at a ratio of 7:3.A nomogram was built based on univariate and multivariate logistic regression analyses in the training group.The discriminatory and predictive capacities of the nomogram were assessed by the receiver operating characteristic curve（ROC）,calibration curve,and decision curves analysis（DCA）.The area under the curve（AUC）was compared by the De Long test.Results A total of 166 lesions were included,of which 45 were benign and 121 were malignant breast lesions.In the training group,univariate analysis showed that age,diameter of lesion,ADC value,BI-RADS,T2_pre,PD_pre,T2_Delta,and T2_ratiowere significantly different between benign and malignant breast lesions（P<0.05）.Multivariate logistic regression analysis showed that ADC value,BI-RADS,and T2_prewere significant variables（all P<0.05）in breast cancer diagnosis.The quantitative model(model A:ADC,T2_pre),BI-RADS model（model B:BI-RADS）,and multi-parameter model(model C:ADC,T2_pre,BI-RADS)were established by combining the above independent variables,among which the diagnostic efficiency of Model A and Model C was significantly higher than that of Model B（the AUC were 0.959,0.976 and 0.810,respectively;Z=3.304,P<0.01;Z=-4.192,P<0.01）.Model A and Model C had the same diagnostic efficiency（Z=-1.338,P=0.181）,the sensitivity and specificity of Model A,Model B,and Model C were 80.00%,96.88%,96.47%,65.63%,91.76%,90.63%,respectively.The calibration curve showed a high agreement between the predicted value of model C with the actual observed values.DCA shows that when the threshold probability ranges between 28%～83%（training group）and 10%～85%（validation group）,the net benefit of model C was higher than that of model A and model B.Conclusion The prediction model established based on DCE-MRI,MUSE,and sy MRI sequence is helpful for clinical differentiation of breast tumors and provides more accurate information for individualized diagnosis.