The diagnostic impact and reproducibility of the different methods use
d within the concept of functional MR-Mammography (FMRM) was assessed
by a multi-reader-analysis. By four experienced readers, 100 histologi
cally confirmed cases were evaluated in six different sessions. Per se
ssion, one of the following components was analyzed: clinical history
(I), static MRM (II), color-coded projection images (III), time-signal
curves of contrast enhancement within a large ROI (IV) and the strong
est enhancing pixel (V) obtained from the histologically confirmed les
ion and the complete FMRM reading (VI). The functional methods (IV-VI)
revealed significantly (p < 0.05) higher specificities than the other
s (I-III). The highest reproducibility between the readers was observe
d for (IV) phi kappa = 0.80, (V) phi kappa = 0.76 and FMRM (VI) phi ka
ppa = 0.63. These three methods also presented the best ROC-curves and
showed the highest complementarity with respect to the false positive
classifications in x-ray mammography. FMRM is a reader independent, r
eproducible method. The analysis of the contrast enhancement time-inte
nsity curves with high temporal resolution allows an improved differen
tiation of malignant and benign findings.