The Value Of Multimodal Magnetic Resonance Functional Imaging And Radiomics In The Evaluation Of Benign And Malignant Breast Lesions

PartⅠThe Study on the value of multimodal functional magnetic resonance imaging in differential diagnosis of benign and malignant breast lesionsObjective:To evaluate the diagnostic efficacy of 3.0T magnetic resonance of IVIM,DKI and DCE-MRI independently and in combination for benign and malignant breast lesions.Methods:Prospective collection of preoperative parameters of MR imaging and dynamic enhanced magnetic resonance imaging and clear the pathologic results of 66 cases of female breast lesions（a total of 71 lesions）,using MITK and obtain a quantitative parameter IVIM DKE software（D,f,D^*）and DKI quantitative parameters（MK,MD）and DCE-MRI parameters(K^rans,K_ep,V_e).The statistical differences of each parameter in benign and malignant breast lesions were analyzed,and the diagnostic efficacy of the parameters with significant differences was evaluated by ROC curve.The parameter with the largest AUC in each scanning mode was selected as the characteristic parameter for joint diagnosis.Results:There were statistically significant differences in D、MD、f、MK、K^trans and K_ep values between benign and malignant breast lesions groups（P<0.05）,but no significant differences in D^*and V_e values（P>0.05）.The AUC values of D、MD、f、MK、K^trans and K_epwere0.940,0.684,0.943,0.848,0.859 and 0.822,respectively,using pathological results as the gold standard.Taking 0.89×10^-3 mm²/s,9.38%,0.75,1.62×10^-3 mm²/s,0.24min^-1and 0.65min^-1 as the optimal diagnostic threshold,the sensitivity were 90.24%,73.17%,92.68%,85.37%,90.14%,78.05%,and the specificity were 96.00%,64.30%,84.00%,80.25%,72.32%,92.20%,respectively.The AUC and sensitivity of the combination diagnosis of the characteristic parameters in any two models were very high,among which the AUC and sensitivity of D combined with Mk value were the highest,which were 0.983 and 99.29%,respectively.Conclusion:IVIM,DKI and DCE-MRI have clinical value for the differential diagnosis of benign and malignant breast diseases.MK value has a large AUC,D value has a high sensitivity and specificity,the combination of the best parameters of any two models has a high AUC and sensitivity,and the combined application of multimodal functional imaging can improve the diagnostic efficiency.PartⅡ The value of multimodal magnetic resonance imaging radiomics in the evaluation of benign and malignant breast lesionsObjective: To investigate the clinical value of multimodal MRI radiomics in the diagnosis of benign and malignant breast lesions.Methods: A retrospective cohort study,collecting 101 cases of breast cancer patients and 44 patients with benign breast lesions of multimodal MR images,manual marking the interested area（ROI）,using Py Radiomics extract image characteristics of omics,by least absolute contraction operator（LASSO）for dimension reduction,screening and benign and malignant diagnosis highest correlation characteristics.A 5-fold cross-validation method was used and a Logistic Regression classifier was used to model the model.The models based on single-mode MR image omics and the comprehensive models based on multimode MR image omics were constructed respectively.Receiver operating characteristic curve（ROC）was used to evaluate the diagnostic performance of the model on the validation set.Results: After dimension-reduction,a total of 30 imaging omics features were left to construct the prediction model based on DCE,DCE0,DWI,T1,T1 FS,T2 and T2 FS single mode MR imaging omics.The AUC in the training set were 0.885,0.790,0.859,0.786,0.793,0.802 and 0.827,respectively.The AUC of the validation set were 0.852,0.736,0.752,0.752,0.750,0.751 and 0.773,respectively.Fusing 17 integrated models from each single modal MR image omics features achieved better performance,with AUC of 0.977 and 0.891 in the training set and the test set,respectively.Conclusion: The AUC of single mode imaging model based on DCE is the largest,and the comprehensive model based on MRI radiomics has a good diagnostic efficiency for benign and malignant breast lesions,and has potential clinical value.