Transformer-based Deep Learning Model Using Contrast-enhanced US for Predicting Malignancy in Breast Nodules: A Two-center Study

Objective To evaluate a novel multichannel deep learning (DL) model using contrast-enhanced ultrasound (CEUS) data with multiple regions of interest (ROIs) and time-intensity curve (TIC)-derived key frames for predicting breast nodule malignancy. Clinical features were integrated into a combined model for robust, generalizable breast lesion classification. Materials and methods This retrospective two-center study enrolled 141 patients with breast nodules: 89 from Institution 1 (June 2016–October 2017; training cohort, n = 62; internal validation, n = 27) and 52 from Institution 2 (November 2022–November 2024; external validation). Tumors were segmented on B-mode and CEUS images to define intratumoral ROIs, tumor bounding boxes, and peritumoral expansions (2 mm and 5 mm). TIC phases (initial, ascending, peak, descending, wash-out) were stacked into multichannel 2.5-dimensional (2.5D) inputs. DenseNet201 models, pretrained on ImageNet, were trained for 2D and 2.5D DL across ROI types. Outputs from the clinical model and optimal intratumoral plus 2-mm peritumoral ROI models were fused via logistic regression. Performance was evaluated using area under the receiver operating characteristic curve (AUC), Hosmer–Lemeshow calibration, decision curve analysis (DCA). Results Among 2.5D models, the multichannel variant with intratumoral plus 2-mm peritumoral ROI showed highest external validation performance. The combined model outperformed individual models externally (AUC 0.949 [95% confidence interval (CI): 0.888, 1.000] vs. clinical AUC 0.821 [95% CI: 0.671, 0.970], p = 0.04; vs. 2D AUC 0.789 [95% CI: 0.660–0.918], p = 0.01; vs. 2.5D AUC 0.824 [95% CI: 0.677, 0.972], p = 0.03). Conclusion This combined model offers promising accuracy and generalizability for CEUS-based breast nodule malignancy prediction, potentially reducing interobserver variability.