Feed

To evaluate the image quality and diagnostic performance of a newly developed three-dimensional automated breast ultrasound (ABUS) system, MammouS-N, characterized by mammography-like geometry. In this prospective study of 121 women with 139 breast lesions, MammouS-N images were obtained in three different projections and reviewed in consensus by two breast radiologists blinded to prior imaging and pathology. Image quality was rated for six parameters. Lesion detectability and BI-RADS categories were compared between MammouS-N and hand-held ultrasound (HHUS), and their concordance was assessed using κ statistics. Diagnostic performance was assessed using the receiver operating characteristic (ROC) analysis with pathology as the reference standard. MammouS-N achieved diagnostically acceptable image quality (score ≥ 4) in over 90% of cases and near-complete coverage in 98.6% of breasts. Of 139 lesions, 131 (94.2%) were visualized, including 88 of 92 malignant lesions (95.7%). The area under the ROC curve for malignancy detection was 0.893 for MammouS-N (95% CI, 0.844-0.943) and 0.923 (95% CI, 0.885-0.962) for HHUS, without a significant difference (p = 0.207). BI-RADS concordance between MammouS-N and HHUS showed moderate-to-substantial agreement (Cohen's κ = 0.45; weighted κ = 0.65), with a mild benign-leaning tendency, as downgrades exceeded upgrades (2.1% of benign and 12.0% of malignant lesions were downgraded relative to HHUS. MammouS-N demonstrated high image quality, excellent lesion detectability, and comparable diagnostic concordance with HHUS. Its mammography-aligned geometry facilitates spatial correspondence across modalities, supporting its potential as a complementary imaging tool for breast cancer screening.