Analysis of 107 breast lesions with automated 3D ultrasound and comparison with mammography and manual ultrasound

Abstract

Objectives Our aim was to investigate the diagnostic potential of an automated ultrasound (US) breast scanner prototype and compare it with manual US and mammography. Methods Ninety-seven patients with a total of 107 breast lesions had mammograms, manual US and an automated breast US scan. Multiplanar reconstructions in coronal, axial and the sagittal view were reconstructed from the automated dataset and visualized. After biopsy, all lesions were confirmed histologically. The data were evaluated according to the BIRADS (Breast Imaging Reporting and Data System) classification. The sensitivity and specificity were analyzed. Results The BIRADS criterion “margin” was significantly related to the overall BIRADS classification, independently of the US method being used. The sensitivity of mammography was significantly lower than of each US method (Fisher's exact test with p < 0.05). There were no significant differences between the US methods. Conclusions The reconstructed third (axial) image plane of the whole breast, which corresponds to a craniocaudal mammogram, can give additional information about both, site and differential diagnosis of a lesion. Although image quality was sufficient, automated US is not good enough to replace manual US at this time.