Diagnostic Efficacy Of Multi-mode Ultrasound In BI-RADS 4 Breast Neoplasms And Establishment Of Predictive Model

Purpose:The aim of this study is to explore the diagnostic efficacy of conventional ultrasound(US)combined with shear-wave elastography(2D-SWE and 3D-SWE)and contrast-enhanced ultrasound(CEUS)in BI-RADS 4 breast neoplasms,and develop a risk prediction nomogram based on the optimal combination to provide reference for clinical management of BI-RADS 4 breast neoplasms.Methods:From September 2020 to June 2021,104 breast neoplasms were categorized as BI-RADS 4 by conventional ultrasound were included in this study(Of the 101 patients,3 had a BI-RADS4 neoplasm on both breast sides).All neoplasms were examined by US,2D-SWE,3D-SWE and CEUS.78 breast neoplasms were randomly assigned into training cohort,compared and analyzed the difference between the size of the region of interest and the different slices in 2D-SWE and 3D-SWE measurement,the diagnostic efficacy of qualitative and quantitative analysis of 2D-SWE,3D-SWE and CEUS,and the diagnostic efficacy of all three individually and in combination on neoplasms were compared,such as the area under the curve(AUC)value,95% confidence interval(95% CI),sensitivity,specificity,positive predictive value(PPV)and negative predictive value(NPV)of 2D-SWE,3D-SWE,CEUS and their combination were compared and analyzed.The optimal combination was selected to screen variables by lasso regression,and developed a risk prediction.The performance of nomogram was assessed by validation cohort of 26 neoplasms.Pathology were confirmed by biopsy.Results:Of the 78 breast neoplasms in the training cohort,16 neoplasms were malignant and 62 neoplasms were benign.There was no significant difference in the size of the region of interest and the different slices for SWE measurement,and there was no significant difference in the qualitative and quantitative analysis of 2D-SWE,3D-SWE and CEUS.After comparison,the performance of US+2D-SWE+CEUS(AUC: 0.986,95% CI: 0.964 to 1.000,Sensitivity: 93.75,Specificity: 91.94)was the best.The nomogram was based on the US+2D-SWE+CEUS(AUC: 0.974,95% CI: 0.911 to 0.997,Sensitivity: 81.25,Specificity: 98.39,PPV: 92.9,NPV: 95.3).The performance of the model was better than radiologists(AUC: 0.863,95% CI: 0.766 to 0.930,Sensitivity: 100.00,Specificity:72.58 PPV: 48.5,NPV: 100.0),there was a significant statistical difference(P<0.001).Among the 26 breast neoplasms in the validation cohort,6 were malignant and 20 were benign.Nomogram showed excellent predictive power in the validation cohort(AUC: 0.946,95% CI:0.780 to 0.997,Sensitivity: 83.33,Specificity: 90.00,PPV: 71.4,NPV:94.7),there was no significant difference in diagnostic accuracy between nomogram and experienced radiologist(AUC: 0.842,95% CI: 0.646 to0.954,Sensitivity: 83.33,Specificity: 85.00,PPV: 62.5,NPV: 94.4).Conclusions:The application of 2D-SWE,3D-SWE and CEUS alone or in combination can improve the diagnostic efficacy of US in BI-RADS 4breast neoplasms.SWE and CEUS showed great complementarity.The diagnostic efficiency of US+3D-SWE was not better than US+2D-SWE.The performance of US+2D-SWE+CEUS was the best,the results showed the nomogram has great diagnostic efficiency and accuracy comparable to that of experienced doctors.The model has high clinical application value.