Diagnostic Performance of Conventional Ultrasound Combined With Calcification Parameters in Soft Tissue Masses With Calcification: A Retrospective Study

ABSTRACT Objective This study aimed to develop and validate an ultrasound‐based model integrating conventional ultrasound and calcification patterns (US‐CAL) for distinguishing benign from malignant calcified soft tissue masses. Methods This retrospective study analyzed 101 pathologically confirmed soft tissue masses (78 benign, 23 malignant) that showed calcification on ultrasound. We developed two models: the US‐CAL model (combining conventional and calcification features) and a conventional ultrasound (US) model. Their performance was compared using receiver operating characteristic (ROC) and decision curve analysis. Results The US‐CAL model outperformed the US model, with a significantly higher area under the curve (AUC) (0.91 [95% CI: 0.85–0.97] vs. 0.85 [0.77–0.93]; p < 0.05). It also demonstrated superior sensitivity (0.91 vs. 0.83), specificity (0.73 vs. 0.71), positive predictive value (PPV) (0.50 vs. 0.45), and negative predictive value (NPV) (0.97 vs. 0.93). DCA confirmed a greater net clinical benefit for the US‐CAL model. Conclusion The US‐CAL model significantly improves the diagnostic accuracy for calcified soft tissue masses, offering a clinically valuable tool to enhance early detection and diagnostic confidence.