Radiomics Analysis Based on Automatic Image Segmentation of DCE-MRI for Predicting Triple-Negative and Nontriple-Negative Breast Cancer

Abstract

Purpose. To investigate whether quantitative radiomics features extracted from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) could be used to differentiate triple-negative breast cancer (TNBC) and nontriple-negative breast cancer (non-TNBC). Materials and Methods. This retrospective study included DCE-MRI images of 81 breast cancer patients (44 TNBC and 37 non-TNBC) from August 2018 to October 2019. The MR scans were achieved at a 1.5 T MR scanner. For each patient, the largest tumor mass was selected to analyze. Three-dimensional (3D) images of the regions of interest (ROIs) were automatically segmented on the third DCE phase by a deep learning segmentation model; then, the ROIs were checked and revised by 2 radiologists. DCE-MRI radiomics features were extracted from the 3D tumor volume. The patients were randomly divided into training ( N = 57 ) and test ( N = 24 ) cohorts. The machine learning classifier was built in the training dataset, and 5-fold cross-validation was performed on the training cohort to train and validate. The data of the test cohort were used to investigate the predictive power of the radiomics model in predicting TNBC and non-TNBC. The performance of the model was evaluated by the area under receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Results. The radiomics model based on 15 features got the best performance. The AUC achieved 0.741 for the cross-validation, and 0.867 for the independent testing cohort. Conclusion. The radiomics model based on automatic image segmentation of DCE-MRI can be used to distinguish TNBC and non-TNBC.

Predicting IDH Mutation Status in Low-Grade Gliomas Based on Optimal Radiomic Features Combined with Multi-Sequence Magnetic Resonance Imaging

“…Recent studies have demonstrated that segmentation repeatability is essential in terms of feature stability, for it is heavily influenced by different MRI protocols and machines [ 21 , 33 , 34 , 35 , 36 ]. In this paper, we downloaded the segmented data outlined on the TCIA website by using automatic image segmentation and manual supervision, and we performed the experiment using the FAE software, which is a publicly available tool for radiomics models and is applied to many fields [ 37 , 38 , 39 , 40 , 41 ]. Thus, all the experiment results are robust and replicable.…”

Development and validation of a deep learning model for breast lesion segmentation and characterization in multiparametric MRI

“…Currently, DCE is most commonly used in AI diagnostic models ( 33 , 35 , 61 , 62 ). The addition of other sequences, such as DWI, to obtain higher diagnostic specificity has also started to be explored ( 63 – 65 ).…”

Radiogenomics, Breast Cancer Diagnosis and Characterization: Current Status and Future Directions

“…Different works were devoted in the last decades to exploring the use of radiomic IFs for different clinical issues. Recently, several studies have explored the application of advanced machine learning methods and deep learning to breast MRI (e.g., [ 60 , 61 ]) publishing a clinical–radiomics model combining a DCE-based radiomics signature and clinical data to predict complete response after neoadjuvant chemotherapy in patients with axillary lymph node metastasis.…”