Imageomics Study On The Identification Of Benign And Malignant Breast Tumors Based On MRI Dynamic Contrast-enhanced Imaging

Purpose:To evaluate a radiomics model of Breast Imaging Reporting and Data System(BI-RADS)3 and 4 breast lesions extracted from breast-tissue dynamic contrast enhanced magnetic resonance(DCE-MR)imaging for lesion characterization by using a sensitivity threshold similar to that of biopsy.Materials and Methods:This institutional study included 317 women at two independent study sites(site 1:267 patients;136 malignant and 181 benign lesions;site 2:50 patients;33 malignant and 17 benign lesions).All women presented with a finding suspicious for cancer at x-ray mammography(BI-RADS 3 or 4)and an indication for biopsy.Before biopsy,DCE-MR imaging was performed by using 3.0-T imagers from different MR imaging vendors.Lesions were segmented and divided into two groups,training group and test group randomly.Three feature model was developed,radiomics clinical model and combined model.The fixed model was tested on an independent test set.And the sensitivities and specificities were evaluated.Results:Three clinical features such as age have a certain correlation with benign and malignant of breast masses.All the three feature models were independently tested.The sensitivity of the training group was 88.4%,the specificity was 68.5%,the sensitivity of the test group was 68.0%,and the specificity was 84.2%.Combined with the clinical features of the imaging ensemble model,the sensitivity of the training group was 72.1%,the specificity was 83.9%,the sensitivity of the test group was 80%,and the specificity was 84.2%.Conclusion:A radiomics model based on DCE-MRI performed by using MR imaging machines from different vendors allowed for reliable differentiation between malignant and benign breast lesions in both a training and an independent test data set.