Differential Diagnosis of Dynamic Contrast-Enhanced-MRI-Based Radiomics Model for Granulomatous Mastitis and Breast Cancer
10.3969/j.issn.1005-5185.2024.02.007
- VernacularTitle:基于动态增强磁共振成像的影像组学模型对肉芽肿性乳腺炎与乳腺癌的鉴别诊断价值
- Author:
Peng LIU
1
;
Xiaojing YU
;
Chunzhi LI
;
Hua REN
;
Yulian MENG
Author Information
1. 中国中医科学院西苑医院放射科,北京 100091
- Keywords:
Breast neoplasms;
Granulomatous mastitis;
Magnetic resonance imaging;
Radiomics;
Diagnosis,differential
- From:
Chinese Journal of Medical Imaging
2024;32(2):144-149
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To investigate the value of a radiomics model based on dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)in the differential diagnosis of granulomatous mastitis and breast cancer.Materials and Methods The MRI data of 82 patients with granulomatous mastitis or breast cancer confirmed by pathology in Xiyuan Hospital of China Academy of Chinese Medical Sciences from February 2019 to January 2022 were retrospectively collected.Based on the first phase of DCE-MRI,the regions of interest(ROI)were delineated layer by layer by semi-automatic segmentation method and manual segmentation method,respectively.99 ROI were randomly assigned to 69 in training groups and 30 in test groups.The consistency difference between the two methods was compared.The original data extracted by the semi-automatic segmentation method were screened by correlation analysis and multi-factor Logistic regression.Six kinds of classifiers(Logistic regression,support vector machine,naive Bayes,decision tree,random forest,K nearest neighbor)were used to construct prediction models,and the differences in diagnostic efficiency,accuracy,sensitivity and specificity of each model were evaluated.Results A total of 99 lesions(n=37 cases with granulomatous mastitis and n=62 cases with breast cancer)were segmented from 82 patients.The radiomics data extracted by the two ROI segmentation methods had poor consistency between groups[Intraclass correlation coefficient=0.68(0.51,0.78)].Among the six prediction models constructed from the data extracted by the semi-automatic segmentation method,the diagnostic performance of the Logistic regression model and the support vector machine model was significantly better than those of other models,and the Logistic regression model had the best diagnostic performance and stability(training group:area under the curve 0.928,accuracy rate 0.855,sensitivity 0.837,specificity 0.885;test group:area under the curve 0.933,accuracy 0.833,sensitivity 0.895,specificity 0.727,respectively).Conclusion Radiomics based on DCE-MRI can provide high value for the differential diagnosis of granulomatous mastitis and breast cancer.The semi-automatic segmentation method is more recommended for the segmentation method of ROI.The prediction model constructed by Logistic regression and support vector machine shows better diagnostic efficiency and stability.
