Can diffusion tensor anisotropy indices assist in breast cancer detection?

PurposeTo evaluate whether the various anisotropy indices derived from breast diffusion tensor imaging (DTI) can characterize the healthy breast structure and differentiate cancer from normal breast tissue. Materials and MethodsSix healthy volunteers and retrospectively selected 24 breast cancer patients were imaged at 3T. DTI included two b-values 0 and 700 sec/mm(2) with 20-64 gradient directions and TE of 120 or 90 msec. The normalized anisotropy indices: fractional anisotropy (FA), relative anisotropy (RA), and 1-volume ratio (1-VR), as well as the absolute maximal anisotropy index ((1)-(3)) were compared. ResultsThe spatial distribution of the various anisotropy indices in healthy volunteers exhibited a high congruence (Pearson correlation coefficients range: 0.79-1.0). All indices showed a statistically significant reduction (P < 0.001) following shortening of the diffusion time. Significantly lower (1)-(3) values were found in cancers as compared to normal breast tissue (P < 6.0 x 10(-7)), while the values of the normalized indices in cancers were not significantly different from those in normal breast tissue (P < 0.65 for FA, P < 0.6 for RA, and P < 0.2 for 1-VR). The contrast-to-noise ratio of (1)-(3) was significantly higher (P < 0.001) than those of the normalized anisotropy indices, and the area under the curve in a receiver operating characteristic analysis exhibited the highest value for (1)-(3) (0.89 0.04 vs. 0.51-0.54 for the other anisotropy indices). ConclusionWater diffusion anisotropy in the healthy breast can be similarly mapped by the normalized indices and by (1)-(3). However, the normalized anisotropy indices fail to differentiate cancer from normal breast tissue, whereas (1)-(3) can assist in differentiating cancer from normal breast tissue. J. Magn. Reson. Imaging 2016;44:1624-1632.