Nomogram model construction and validation of integrating ultrasonic and mammographic features for diagnosing benign and malignant breast lesions

Background Breast cancer remains a leading global health concern, emphasizing the need for accurate diagnostic strategies. This study aimed to develop and validate a nomogram model integrating ultrasonographic and mammographic features for discriminating between benign and malignant breast lesions. Methods A retrospective, multi-center cohort study included 407 female patients with breast lesions who underwent both ultrasound and mammography between October 2022 and September 2025. Patients from two hospitals formed the training cohort (n = 285) and those from a third independent hospital formed the validation cohort (n = 122), yielding an approximate 7:3 ratio. Key ultrasonographic and mammographic features, defined by the Breast Imaging Reporting and Data System (BI-RADS) lexicon, along with clinical factors, were analyzed. Feature selection was performed using LASSO regression, followed by multivariable logistic regression to construct the nomogram. Results In the training cohort (153 benign, 132 malignant), the integrated nomogram model achieved an area under the curve (AUC) of 0.937 (95% CI: 0.884-0.991), with a sensitivity of 0.916 and specificity of 0.892, outperforming ultrasound-only (AUC = 0.872) and mammography-only (AUC = 0.891) models. Key predictors included age ≥51.50 years, diabetes, larger tumor diameter, postmenopausal status, and specific BI-RADS features from both modalities. In the independent validation cohort (66 benign, 56 malignant), the model maintained robust performance with an AUC of 0.934 (95% CI: 0.899-0.969), a sensitivity of 0.911, and a specificity of 0.894. Conclusion The developed nomogram effectively integrates multimodal imaging and clinical features to preoperatively estimate the probability of malignancy in breast lesions, demonstrating high diagnostic accuracy and potential clinical utility.