Figure 2.

Set of diffusion tensor distributions (DTDs) for individual voxels extracted from a T2-weighted non diffusion-weighted S₀ map with an invasive ductal carcinoma in the right breast. The distributions are represented in two-dimensional plots of diffusion tensor sizes (D_iso) and shapes (D_∆²). Size (D_iso) reports on cellularity in an inversely proportional relation. Shape (D_∆²) values range from 0 for spherical tensors to 1 for elongated tensors. Voxels (A,B) including cancer exhibit diffusion in densely packed elongated cells and tumoral isotropic cells. Voxel (C) presents isotropic cells with higher diffusivity corresponding to healthy fibroglandular tissue. The directional color-coding is based on the diffusion tensor eigenvalues, normalized by the maximum eigenvalue, and reports on the orientation of the underlying diffusion patterns. The red/green/blue directions correspond to the left-right/anterior-posterior/superior-inferior directions, respectively. The key diffusion properties of these cellular configurations can be quantified via the statistical descriptors of the DTD, i.e., means and (co)variances calculated over the size (D_iso), shape (D_∆²), and orientation dimensions of the DTD. The diffusion encoding strategies offered by multidimensional diffusion (MDD) acquisitions enable simultaneous measurement of various features of the DTD, thereby enhancing the specificity of the microstructural information arising from diffusion processes in vivo.