Automated breast ultrasound (ABUS) meets artificial intelligence: a new accurate classification tool for BIRADS 3-4 breast lesions.

Authors

Marco Moschetta, Alessia Spitaleri, Federico Cofone, Fabio Panarelli, Corrado Caiazzo, Amato Antonio Stabile Ianora, Michele Telegrafo

Published in

Journal of ultrasound. Aug 09, 2026. Epub Aug 09, 2026.

Abstract

Breast ultrasound (US) has often interpretation challenges (BIRADS 3-BIRADS 4 lesions), leading to a high demand for core needle biopsies (CNB). In this context, a computer-aided diagnostic (CAD) tool based on a machine-learning artificial intelligence, applied to images obtained with an automated breast ultrasound system (ABUS), could potentially provide more accurate diagnoses by instantly and automatically assessing the US lesion features and their level of suspicion. The aim of this study is to evaluate the diagnostic accuracy of a CAD applied to ABUS images for BIRADS 3-4 lesions.
Fifty-four patients underwent ABUS, performed with three scans per side, reconstructing images in the axial, coronal, and sagittal planes. Two expert radiologists draw a region of interest (ROI) on the lesion to be assessed by CAD tool, reaching a consensus regarding the images. Only breast lesions classified as BIRADS 3-4 were included, for a total of 68 (32 BIRADS 3, 36 BIRADS 4). CAD results were compared with the histological findings from CNBs in all cases, calculating sensitivity (SE), specificity (SP), accuracy (AC), positive predictive value (PPV), and negative predictive value (NPV).
Histology identified 40 positive lesions and 28 negative lesions. The CAD classified 38 suspicious lesions, with 6 false positives, 22 true negatives, and 2 false negatives. The values for SE, SP, AC, PPV, and NPV were 95, 79, 88, 86, and 92%, respectively.
The proposed CAD system is accurate in characterizing BIRADS 3-4 breast lesions diagnosed by ABUS and, therefore, could be applied in the clinical practice in order to assist radiologists in evaluating US images and reducing the number of the required CNBs.

PMID:
42572099
Bibliographic data and abstract were imported from PubMed on 10 Aug 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.