The Value Of X-ray And MRI Imaging Features With X-ray Radiomics In Differentiating Benign And Malignant Breast Phyllodes Tumor

Part 1 The value of differentiating benign and malignant breast phyllodes tumor based on X-ray and MRI imaging featuresObjective To analyze the X-ray and Magnetic Resonance Imaging(MRI)characteristics of breast phyllodes tumor and evaluate the value of differentiating benign and malignant breast phyllodes tumor based on X-ray and MRI imaging features.Material and Methods The clinical,pathological,and imaging data of 80 patients with phyllodes tumor diagnosed by pathology in Zhongda Hospital affiliated to Southeast University from January 2007 to November 2021 were retrospectively analyzed.Among them,76 patients underwent mammography and 34 patients underwent MRI.According to pathological findings,patients were divided into benign phyllodes tumor group(benign group)and borderline/malignant phyllodes tumor group(malignant group).According to the standard of American College of Radiology(ACR)breast Imaging Reporting and Data System(BI-RADS),X-ray features include maximum diameter,shape,leaf edge points,density,internal calcification,breast density,and "halo sign".MRI features include maximum diameter,morphology,margin,leaf,T2 WI lipid compression signal,T1 WI hyperintensity signal,presence or absence of cystic components,internal separation,time-signal intensity curve(TIC),and mode of enhancement.Independent sample T-test was used for measurement data of normal distribution,and the Mann-Whitney U test was used for measurement data of skewness distribution.The image features between the two groups were analyzed by the Chi-square test or Fisher’s exact probability method.The predictive model was constructed by Logistic regression,and the diagnostic efficiency of the image feature model was evaluated by receiver operating characteristic curve(ROC)and Area under curve(AUC).Results The mean age of the benign group was 41.25±9.80 years,and that of the malignant group was 50.00±12.25 years,the difference between the two groups was statistically significant(P=0.001).The maximum diameter in X-ray features was significantly different between the two groups(P=0.003),3.1(2.3 ~ 4.8)cm in the benign group and 5.0(3.1 ~ 8.1)cm in the malignant group.The AUC value of the combined model based on age and the maximum diameter of lesions was 0.777,and the sensitivity and specificity were 0.667 and 0.776,respectively.The T1 WI high signal in MR characteristics was significantly different between the two groups(P=0.034),and the TIC curve type was significantly different between the two groups(P=0.010).The AUC value of TIC curve type and T1 WI high signal combined model was 0.796,and the sensitivity and specificity were 0.500 and 0.917,respectively.Conclusion Partial imaging features of X-ray and MRI can be used to distinguish benign and malignant breast phyllodes tumor before surgery.Part 2 The value of differentiating benign and malignant breast phyllodes tumor based on X-ray radiomic featuresObjective To evaluate the value of differentiating benign and malignant breast phyllodes tumor based on X-ray radiomic features.Material and Methods The imaging data of 63 patients who were pathologically diagnosed with phyllodes tumor and underwent mammography from January 2007 to November 2021 in Zhongshan Hospital affiliated to Southeast University were retrospectively analyzed.The patients were divided into benign phyllodes tumor group(benign group)and borderline/malignant phyllodes tumor group(malignant group).The Region of interest(ROI)was delineated by manual segmentation method,the F test was used for feature screening,the prediction model was established by Support Vector Machine(SVM),and the model verification was randomly split by internal verification.70% were training sets(44 cases)and 30% were validation sets(19 cases).The diagnostic efficiency of the model was evaluated by ROC curve and AUC value.Results A total of 900 image omics features were extracted,and 10 features were screened out through the F test to construct the classifier model.The results showed that the imaging omics model based on X-ray photography had good performance in differentiating the benign and borderline/malignant phyllodes tumor.The AUC value of the training set was 0.94,and the sensitivity,specificity,and accuracy were 0.89,0.85,and 0.86.The AUC value of the validation set was 0.87,and the sensitivity,specificity,and accuracy were 0.75,0.80,and 0.78.Conclusion The radiomic model based on mammography has a high value in differentiating benign and malignant phyllodes tumor.