Hormone receptor (HR) status is a critical biomarker used to formulate treatment programs and prognosis in breast cancer. Traditional immunohistochemistry relies on invasive tissue samples and may not accurately reflect tumor heterogeneity. Radiomics is a noninvasive technique that involves extracting quantitative imaging characteristics from molecular profiles. The purpose of this study is to develop a combined ultrasound (US)-radiomics model to predict HR status in invasive breast cancer. A retrospective cohort of 186 patients with invasive breast carcinoma, which had been pathologically confirmed, was used in this study, comprising 150 cases (HR-positive (ER + /PR − , HER-2−)) and 36 cases (HR-negative (ER − /PR − , HER-2−)). B-mode US images of the tumor regions were manually segmented, and 463 radiomic features were obtained. T tests, ANOVA, and recursive methods were applied to create a list of features. A support vector machine with a radial basis function kernel was trained using leave-one-out cross-validation. To measure model performance, accuracy, sensitivity, specificity, area under the curve (AUC), and 95% confidence intervals (CIs) were used. The hybrid model had an AUC of 0.728 (95% CI: 0.701–0.755) and an accuracy of 67.9. The model with the highest AUC (0.753, 95% CI: 0.7240.782) was the internal echo-based model. HR-negative tumors were larger, had higher marker of proliferation (Ki-67) indices, and showed greater textural heterogeneity than HR-positive lesions ( P < .05). US-radiomics combined modeling is a promising, cost-effective, and radiation-free approach to noninvasive imaging that can predict HR status in breast cancer. US biomarkers can be quantitative, providing insights into tumor microstructure to personalize diagnostic and therapeutic approaches.
Xue et al. (Fri,) studied this question.