Prospective Evaluation of Super-Resolution Ultrasound for Differentiating Benign and Malignant Small Breast Nodules

Objective: The objectives of this study were to identify key super-resolution ultrasound (SR US) microvascular parameters via cluster analysis and evaluate their value in differentiating benign from malignant breast nodules of size ≤2 cm, and to assess the added value of integrating conventional US features. Methods: This study screened 481 breast nodules by conventional and SR US; 123 nodules of size ≤2 cm were included. Spearman hierarchical clustering, within-cluster principal component analysis, and iterative variance inflation factor elimination selected core parameters. With histopathology as the reference, five models were built: conventional US morphological features (M1), quantitative SR US (M2), M2 plus enlarged enhancement area and penetrating vessels (M3), Model 1 plus enlarged enhancement area and penetrating vessels (M4), and all features (M5). Performance was compared by AUCs and the DeLong test, with bootstrap optimism correction, Hosmer–Lemeshow calibration, and decision curve analysis (DCA). Results: Five parameters were retained: vascular density (VD), fractal dimension (FD), mean intensity (Mean Inten), standard deviation of velocity (Std Vel), and minimum velocity (Min Vel). The AUCs for M1-M5 were 0.702, 0.752, 0.820, 0.793, and 0.852, and their sensitivity, specificity, positive predictive value, and negative predictive value (%) were 50.0/82.2/65.8/70.6, 58.0/82.2/69.0/74.1, 86.0/65.8/63.2/87.3, 80.0/69.9/64.5/83.6, and 88.0/71.2/67.7/89.7, respectively; M5 performed best (AUC 0.852; sensitivity 88.0%, specificity 71.2%, PPV 67.7%, NPV 89.7%). The bootstrap-corrected AUCs were 0.684, 0.716, 0.779, 0.732 and 0.799; calibration was adequate and DCA confirmed a net benefit for SR US models. Conclusions: VD, FD, Mean Inten, Std Vel, and Min Vel differentiate benign from malignant breast nodules of size ≤ 2 cm. SR US outperforms conventional US, and integrating morphologic features further improves discrimination.