Development and validation of a screening model for benign and malignant breast masses based on S-Detect and microvascular flow imaging.

Background: Imaging examination of a breast mass is essential for improving breast cancer detection. Previous screening models of benign and malignant breast masses demonstrated a high level of subjectivity due to the inability to conduct quantitative evaluations. Thus, this study aimed to construct an objective, convenient, and effective nomogram incorporating S-Detect and microvascular flow imaging (MVFI) to predict breast cancer risk.

Methods: Female patients with breast masses detected by conventional ultrasound examinations at the Second Affiliated Hospital of Anhui Medical University between January 2021 and October 2024 were retrospectively analyzed. All patients underwent preoperative assessments with both S-Detect and MVFI. The pathological results served as the gold standard for diagnosis. After screening, a total of 724 breast masses from 712 patients were randomized into the training (506 masses) and validation (218 masses) groups. Univariate analysis assessed patient age, as well as the location, size, vascular index (VI), and S-Detect-based diagnosis of the masses. Risk factors for predicting breast cancer were screened using multivariate analysis. A nomogram prediction model was then constructed. Diagnostic performance, clinical utilization value, and calibration were determined using the receiver operating characteristic (ROC) curve, decision curve analysis (DCA), and calibration curve, respectively. Nomogram risk was calculated for each breast mass for risk stratification.

Results: The training group included 208 benign and 298 malignant masses, while the validation group comprised 85 benign and 133 malignant masses. Multivariate analysis demonstrated that mass size [odds ratio (OR) =1.08; P<0.001], age (OR =1.09; P<0.001), VI (OR =1.07; P<0.001), and S-Detect-based diagnosis (OR =28.37; P<0.001) were risk factors for predicting breast cancer. The area under the curve (AUC) for the nomogram model was significantly greater than that for S-Detect in both the training (0.93 vs. 0.82, P<0.001) and validation (0.91 vs. 0.82, P<0.001) groups. The diagnostic sensitivity and specificity of the nomogram were 93.3% and 79.8% in the training group, and 98.5% and 72.9% in the validation group, respectively. The optimal cut-off value for nomogram risk differentiation between the high-risk and low-risk sets was 0.495, with a significantly higher proportion of malignant breast masses in the high-risk set compared to that in the low-risk set (P<0.001).

Conclusions: This novel nomogram model based on quantitative and objective ultrasound and clinical features can quantify the malignancy risk of breast masses, identify high-risk individuals, and provide a reference for further examinations.