Size assessment of breast lesions by means of a computer-aided detection (CAD) system for magnetic resonance mammography.

ABSTRACT

PURPOSE: The aim of this study was to investigate the efficacy of a dedicated software tool for automated volume measurement of breast lesions in contrast-enhanced (CE) magnetic resonance mammography (MRM). MATERIAL AND METHODS: The size of 52 breast lesions with a known histopathological diagnosis (three benign, 49 malignant) was automatically evaluated using different techniques. The volume of all lesions was measured automatically (AVM) from CE 3D MRM examinations by means of a computer-aided detection (CAD) system and compared with the size estimates based on maximum diameter measurement (MDM) on MRM, ultrasonography (US), mammography and histopathology. RESULTS: Compared with histopathology as the reference method, AVM understimated lesion size by 4% on average. This result was similar to MDM (3% understimation, not significantly different) but significantly better than US and mammographic lesion measurements (24% and 33% size underestimation, respectively). CONCLUSIONS: AVM is as accurate as MDM but faster. Both methods are more accurate for size assessment of breast lesions compared with US and mammography.