Texture Analysis of Breast DCE-MRI Based on Intratumoral Subregions for Predicting HER2 2+ Status

Abstract

Background: Breast tumor heterogeneity is related to risk factors that lead to aggressive tumor growth; however, such heterogeneity has not been thoroughly investigated.Purpose: To evaluate the performance of texture features extracted from heterogeneity subregions on subtraction MRI images for identifying human epidermal growth factor receptor 2 (HER2) 2+ status of breast cancers.Materials and Methods: Seventy-six patients with HER2 2+ breast cancer who underwent dynamic contrast-enhanced magnetic resonance imaging were enrolled, including 42 HER2 positive and 34 negative cases confirmed by fluorescence in situ hybridization. The lesion area was delineated semi-automatically on the subtraction MRI images at the second, fourth, and sixth phases (P-1, P-2, and P-3). A regionalization method was used to segment the lesion area into three subregions (rapid, medium, and slow) according to peak arrival time of the contrast agent. We extracted 488 texture features from the whole lesion area and three subregions independently. Wrapper, least absolute shrinkage and selection operator (LASSO), and stepwise methods were used to identify the optimal feature subsets. Univariate analysis was performed as well as support vector machine (SVM) with a leave-one-out-based cross-validation method. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate the performance of the classifiers.Results: In univariate analysis, the variance from medium subregion at P-2 was the best-performing feature for distinguishing HER2 2+ status (AUC = 0.836); for the whole lesion region, the variance at P-2 achieved the best performance (AUC = 0.798). There was no significant difference between the two methods (P = 0.271). In the machine learning with SVM, the best performance (AUC = 0.929) was achieved with LASSO from rapid subregion at P-2; for the whole region, the highest AUC value was 0.847 obtained at P-2 with LASSO. The difference was significant between the two methods (P = 0.021). Conclusion:The texture analysis of heterogeneity subregions based on intratumoral regionalization method showed potential value for recognizing HER2 2+ status in breast cancer.

Radiomic Analysis of Pharmacokinetic Heterogeneity Within Tumor Based on the Unsupervised Decomposition of Dynamic Contrast‐Enhanced MRI for Predicting Histological Characteristics of Breast Cancer

“…This is consistent with previous findings. 38,39 The results indicate that the 3D tumor subregion related to FFK is closely associated with tumor growth and angiogenesis, 25 and thus can demonstrate better diagnostic performance for molecular subtypes, Ki-67, and tumor grade. Our results also showed lower diagnostic performance for the basal-like subtype than that of the other subtypes.…”

“…Previous studies, which focused on the whole tumor, have identified imaging biomarkers associated with histological characteristics 34–36 . Radiomic analysis based on intra‐tumoral regions showed that features from the subregions were closely associated with histological characteristics than those based on the whole tumor 25,37,38 . Different from these analyses of intra‐tumoral heterogeneity, this study investigated PK heterogeneity within tumor for enhancing the prediction of molecular subtypes, Ki‐67, and tumor grade.…”

“…[34][35][36] Radiomic analysis based on intratumoral regions showed that features from the subregions were closely associated with histological characteristics than those based on the whole tumor. 25,37,38 Different from these analyses of intra-tumoral heterogeneity, this study investigated PK heterogeneity within tumor for enhancing the prediction of molecular subtypes, Ki-67, and tumor grade. As a consequence, the tumor subregions obtained better diagnostic performance than the whole tumor.…”

“…Texture features have commonly been used to capture spatial heterogeneity in tumor in previous studies. 8,25,38 In addition to texture features, this study evaluated the diagnostic performance of shape features and statistics features. In the univariate analysis, the individual feature of voxel volume, dependence non-uniformity normalized, and large dependence emphasis from the 3D tumor subregion related to FFK achieved the best diagnostic performance for molecular subtypes, Ki-67, and tumor grade, respectively.…”

Texture Analysis of Breast DCE-MRI to Predict Benign and Malignant Tumors and HER-2 Status Using Machine Learning and Deep Learning

“…Over the course of the last 5-10 years, TA has been increasingly become a tool for maximizing the amount of information extracted from virtually any medical imaging modality [12][13][14]. Least absolute shrinkage and selection operator (LASSO) [15], logistic regression analysis (LRA) [16], and support vector machine (SVM) [17] have become very popular machine learning tools for analyzing complex data that often arise in radiological research. LASSO is a novel feature selection method that can use a penalty function to filter data with a large number of covariates [18][19].…”