The Value Of Radiomics Nomogram Based On Multimodal Magnetic Resonance Imaging In Predicting Sentinel Lymph Node Metastasis In Invasive Breast Cancer

Objective: The nomogram based on multimodal magnetic resonance imaging radiomics was constructed to predict sentinel lymph node metastasis in invasive breast cancer,in order to provide strategies for clinical treatment decisions.Methods: Preoperative breast MRI images and clinical pathological data of 165 breast cancer patients were collected and analyzed.The patients were randomly divided into training set(n=132)and validation set(n=33)at a ratio of 8:2.Radiomics features were extracted from contrast-enhanced T1-weighted image(CE-T1WI),fat-suppressed T2-weighted image(FS-T2WI)and diffusion kurtosis image(DKI).The least absolute shrinkage and selection operator(LASSO)was used to select radiomics features.The support vector machine(SVM)was used to construct radiomics models.Radiomics nomogram was constructed by logistic regression(LR).The diagnostic performances of the models were evaluated by receiver operating characteristic(ROC)curve and area under the curve(AUC).Delong test was used to test whether there were statistical differences between the analysis models.Decision curve analysis(DCA)was used to determine the application value of the nomogram in clinical practice.Results: Fourteen features were selected to construct the final radiomics model that constructed based on CE-T1WI,FS-T2WI and DKI(b=2000).The AUCs of the radiomics model in the training set and the validation set were 0.913(95%CI: 0.810,1.000)and 0.881(95%CI:0.818,0.945),respectively.The clinical model(tumor edge)produced AUCs of 0.726(95%CI: 0.573,0.879)and 0.709(95%CI: 0.645,0.773)in the training and validation set,respectively.In the training and validation sets,the AUCs of the radiomics nomogram model were 0.960(95%CI:0.904,1.000)and 0.935(95%CI: 0.895,0.974),respectively.In the training set and the validation set,the AUC of the radiomics nomogram model was higher than that of the clinical model(P=0.006).There was no significant difference in AUC between the radiomics model and the nomogram model(P=0.162).Conclusion: The multimodal MRI radiomics model and the nomogram model combined with tumor margin in this study are helpful for preoperative diagnosis of SLN metastasis in breast cancer and have a high predictive efficiency.