A feasibility study of classification between breast carcinoma in situ and invasive carcinoma using intratumoral and peritumoral radiomics based on dynamic contrast-enhanced MRI
10.3760/cma.j.cn112149-20211222-01130
- VernacularTitle:瘤内及瘤周动态增强MRI影像组学特征鉴别乳腺原位癌与浸润性癌的可行性研究
- Author:
Yuan JIANG
1
;
Yuanjia CHENG
;
Li GUO
;
Mingming MA
;
Yaofeng ZHANG
;
Xiaodong ZHANG
;
Xiaoying WANG
;
Naishan QIN
Author Information
1. 北京大学第一医院医学影像科，北京 100034
- Keywords:
Breast neoplasms;
Magnetic resonance imaging;
Carcinoma in situ;
Radiomics;
Peritumoral
- From:
Chinese Journal of Radiology
2022;56(9):976-981
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the feasibility of classification between carcinoma in situ and invasive carcinoma of breast using intratumoral and peritumoral radiomics based on breast dynamic contrast-enhanced (DCE) MRI.Methods:The retrospective study included consecutive invasive breast carcinoma pathological diagnosed by core needle biopsy or surgery from January 2013 to December 2013 and carcinoma in situ of breast diagnosed by surgery from January 2013 to December 2015 in Peking University First Hospital. All patients had pretreatment breast MRI images. A total of 251 cases (251 lesions) were included, with 208 invasive breast carcinoma and 43 carcinoma in situ of breast. They were all females and median age was 53 (23-82) years old. Patients were randomly divided into the training ( n=176) and testing dataset ( n=75) in a 7∶3 ratio. In the training dataset, combined with DCE mask and early enhancement images, intratumoral and peritumoral area were semi-automatic segmentation, and radiomics features were extracted and dimension reduction, finally a prediction model was established. Model performance was tested in the testing dataset. Receiver operating characteristic (ROC) curve and area under curve (AUC) were used to analyze the model prediction performance. Results:The prediction models established by intratumoral, peritumoral and intratumoral combined with peritumoral radiomics had good performance. The AUC of intratumoral, peritumoral and intratumoral combined with peritumoral radiomics prediction models in differentiating breast carcinoma in situ and invasive carcinoma were 0.865, 0.896 and 0.922 in the testing dataset, there was no significant difference in pairwise comparisons ( P>0.05). The sensitivity of intratumoral, peritumoral and intratumoral combined with peritumoral radiomics prediction models were 77.4%, 87.1%, 83.9%, the specificity were 92.3%, 84.6%, 100%, and the accuracy were 80.0%, 85.3%, 86.7%. Conclusion:It is potential feasible for classification between carcinoma in situ and invasive carcinoma of breast using intratumoral and peritumoral radiomics based on breast DCE MRI.
