A Multimodal Diagnostic Model for Breast Cancer Invasiveness Based on Ultrasound Imaging and Serum Biomarkers

Background and Objectives: Breast cancer invasiveness significantly impacts treatment strategies and prognosis. Combining ultrasound imaging modalities with serum biomarkers may improve diagnostic accuracy. We aimed to develop and validate a multimodal diagnostic model integrating ultrasound B-mode, Doppler imaging, and serum biomarkers for assessing breast cancer invasiveness. Materials and Methods: A multimodal diagnostic model was developed using ultrasound B-mode, Doppler imaging, and serum biomarkers from patients with invasive and non-invasive breast cancer. Machine learning algorithms, including Logistic Regression, Random Forest, and XGBoost, were applied to predict invasiveness, with performance evaluated using accuracy, precision, recall, F1-score, and AUC. Results: The multimodal model outperformed single-modality approaches, with XGBoost achieving the highest accuracy (88.90%) and AUC (0.930). The inclusion of specific serum biomarkers (e.g., CA125, CA15-3, CEA, and CA19-9) significantly enhanced diagnostic accuracy for breast cancer invasiveness. Conclusions: The proposed multimodal diagnostic model integrating ultrasound imaging and serum biomarkers provides a highly accurate and reliable method for assessing breast cancer invasiveness, offering potential to improve clinical decision-making and patient outcomes.