Outside the screening context, breast lesions categorized as BI-RADS 0 are very rarely encountered because breast MRI is typically performed after conventional mammography and ultrasound (Mendelson et al. 2003).

Criteria used to assess breast lesions on contrast-enhanced MR images include the shape and demarcation of enhancing lesions, the pattern of enhancement, and the time course of enhancement. Malignancy is suggested by an irregular, star-shaped appearance with ring enhancement or rapid initial enhancement with postinitial washout, while a sharply marginated, round, or oval lesion with slow and continuous contrast medium uptake tends to be benign (Fischer 2012; Macura et al. 2006) (Table 11.2).

The so-called Goettingen score has been developed and used as a standardized tool for the comprehensive multifactorial evaluation of breast abnormalities on MRI and is most useful for the characterization of hypervascularized mass lesions. The tool integrates a maximum of criteria that have shown promise in differentiating between malignant and benign breast lesions. The strongest malignancy criteria – ring enhancement, strong initial enhancement (>100 %), and postinitial washout – are each assigned a score of 2. Less suspicious features are assigned a score of 1 and include irregular shape, ill-defined margins, inhomogeneous enhancement, 50–100 % initial contrast enhancement, and postinitial plateau. A score of 0 is assigned to features suggesting a benign lesion; these include round or oval shape, sharp margin, homogeneous enhancement, moderate initial signal increase of <50 %, and continuous postinitial signal increase. A lesion can thus be assigned a total score of 0–8 points. Breast lesions with a total Goettingen score of 1 or 2 (corresponding to ACR BI-RADS categories 1 and 2) are more likely benign, while lesions assigned a score of 4 or higher (corresponding to ACR BI-RADS categories 4 and 5) are more suspicious for malignancy (Fischer 2012).

Additional information for characterizing a breast lesion is provided by postprocessing tools that analyze contrast enhancement kinetics over time. This is done in a region of interest (ROI) placed in an area of contrast enhancement. In case of inhomogeneous lesion enhancement, the ROI is placed in the area showing the strongest enhancement. The ROI should include at least three pixels because misinterpretation of the enhancement pattern may result when the ROI is too small.

The signal increase after contrast medium administration can be expressed as a percentage relative to the baseline value on precontrast images or as a normalized value relative to the signal intensity of surrounding fat.

Strong initial enhancement with rapid washout is characteristic of a malignant breast lesion, while moderate initial enhancement with a slight further increase or plateau phase thereafter favors a benign lesion (Table 11.2).

Pitfalls in image interpretation arise from focal areas of enhancement within the glandular tissue in young premenopausal women (Amarosa et al. 2013) and in postmenopausal women on hormone replacement therapy (Pfleiderer et al. 2004). These foci may show the same rapid and strong initial enhancement as breast cancer. However, these foci are usually temporary and will no longer be detectable in a repeat examination (Kuhl et al. 1995; Liberman et al. 2002).

More recent studies suggest that fat-saturated diffusion-weighted SE images with generation of apparent diffusion coefficient (ADC) maps can help in characterizing breast lesions with inconclusive morphologic MRI features and dynamic kinetic patterns (Rahbar et al. 2011 ; Tsushima et al. 2009). High signal intensity of a suspicious lesion on diffusion-weighted images with low values in ADC maps suggests reduced perfusion and hypercellularity. Benign lesions or tumors that have responded to treatment usually have normal diffusion (De Bazelaire et al. 2010).

The following sections discuss the most important and most common incidental breast findings detected by MR mammography in the setting of whole-body MRI screening.