Compare tomosynthesis to mammography, ultrasound, MRI, and histology for the detection and staging of BI-RADS 4–5 anomalies, as a function of breast composition, lesion location, size, and histology.

Seventy-five patients underwent mammography, tomosynthesis, ultrasound, and MRI. The diagnostic accuracy of the different examinations was compared.

The sensitivities for detection were as follows: 92.5% with MRI, 79% for ultrasound, 75% for tomosynthesis, and 59.5% for mammography. Tomosynthesis improves the sensitivity of mammography (P=0.00013), but not the specificity. The detection of multifocality and multicentricity was improved, but not significantly. Tomosynthesis identified more lesions than mammography in 10% of cases and improved lesion staging irrespective of the density, but was still inferior to MRI. The detection of ductal neoplasia was superior with tomosynthesis than with mammography (P=0.016), but this was not the case with lobular cancer. The visualization of masses was improved with tomosynthesis (P=0.00012), but not microcalcifications. Tomosynthesis was capable of differentiating lesions of all sizes, but the smaller lesions were easier to see. Lesion sizes measured with tomosynthesis, excluding the spicules, concurred with histological dimensions. Spicules lead to an overestimation of the size.

In our series, tomosynthesis found more lesions than mammography in 10% of patients, resulting in an adaption of the surgical plan.