Differential Diagnosis Value Of MRI Multi-Parameter And Texture Analysis In Non-Lactation Mastitis And Breast Cancer

Part 1:Differential diagnosis value of MRI multi-parameter in nonlactation mastitis and breast cancerObjective:To explore the value of MRI multi-parameter in differential diagnosis of non-lactation mastitis and breast cancer.Methods:The clinical and imaging data of 34 cases of non-lactation mastitis and 68 cases of breast cancer in our hospital from January 2016 to December 2020 were retrospectively collected,and analyzed two groups about the qualitative data including lesion type,the edge of lesions,whether the cumulative nipple areola,whether to have local thickening of skin edema,with or without axillary lymph nodes,and lesions of TIC type,will count data SPSS25.0 software is adopted to improve the chi-square test,check whether the difference between has statistical significance.And measured investigate,Emax,Tmax,ADC values,quantitative parameters,such as the quantitative parameter SPSS25.0 software is adopted to improve the t test or nonparametric test,inspection and qualitative and quantitative MRI parameters on whether the difference between the two groups have statistical significance,and MRI qualitative and quantitative parameter binary logistic regression model was constructed,and ROC Medcalc19.3 software analysis,get the AUC,Z values,P values,sensitivity,specific degrees and 95%confidence interval.Results:There were significant differences in lesion morphology,lesion edge and TIC curve type between non-lactation mastitis and breast cancer(P<0.05),and the corresponding AUC were 0.640,0.682 and 0.693.There were significant statistical differences in MRI quantitative parameters EER,EMAX and ADC values between the two groups,with P<0.05 and AUC of 0.745,0.794 and 0.785,respectively.There was no statistically significant difference in Tmax(P=0.8002).Multivariate binary logistic regression analysis was performed for each MRI qualitative parameters,quantitative parameters and quantitative combined qualitative parameters with statistical significance,and the AUC was 0.801,0.824 and 0.890,respectively.Conclusion:MRI qualitative and quantitative parameters have a certain value in the differential diagnosis of non-lactation mastitis and breast cancer.The combination of MRI qualitative and quantitative parameters can further improve the differential diagnosis value of non-lactation mastitis and breast cancer.Part 2:Value of MRI texture analysis in differential diagnosis of non-lactation mastitis and breast cancerPurpose:To explore the value of MRI texture analysis in the differential diagnosis of non-lactation mastitis and breast cancer.Methods:Through analysis T1WI、T2WI、T1WIenhanced phase,DWI and ADC images of 34 cases of non-lactation mastitis and 68 cases of breast cancer confirmed by pathology in our hospital from January 2016 to December 2020 were retrospectively collected.The image was imported into Mazda software in DICOM format,and the region of interest(ROI)was delineated on the maximum cross section of the lesion to extract the texture characteristic parameters.Three different screening methods were used to reduce the dimensionality of 304 original texture parameters extracted.Fisher coefficient(Fisher coefficient),Classification error probability combined with average correlation coefficients(POE+ACC),and mutual information(MI).Each screening method can obtain 10 optimal subsets of texture features.The differences among the three screening methods were analyzed,and the optimal subsets of the three screening methods were used to construct a binary logistic regression model.The area under the curve(AUC),Z value,P value,Youden index,sensitivity,specificity and 95%confidence interval were analyzed.Four feature classification methods including RDA,LDA,PCA and NDA,provided by B11 in Mazda software,were used to calculate the misjudgment rate of texture features.Qualitative and quantitative parameters of MRI were combined with the optimal subset parameters of MRI texture features to analyze the diagnostic efficiency.Results:The texture characteristic parameters had excellent diagnostic efficacy on T2WI and T1WI(AUC=0.926 and 0.911),and had good diagnostic efficacy on TI enhancement phase 5,DWI and ADC sequence(AUC=0.809-0.890).The lowest misjudgment rate with NDA was 15.69%(16/102)under TIWI,TIWI enhanced phase 5 and Fisher’s screening method in DWI sequence.The combination of clinical qualitative and quantitative parameters with the optimal T2WI texture feature subset significantly improved the diagnostic performance(AUC was 0.968).Conclusion:MRI texture analysis has a good value in the differential diagnosis of nonlactation mastitis and breast cancer,especially T2WI sequence is the best.