Characterizing different brain structures based on information content: Diffusion entropy

To overcome errors introduced via the assumption of a Gaussian diffusion tensor model when dealing with voxels having multiple fiber orientations, a new measurement method is introduced to evaluate white matter properties from diffusion-weighted magnetic resonance imaging using an entropy concept. Here, the entropy results were compared, qualitatively and quantitatively, with Fractional Anisotropy (FA), and were validated via histology evaluation of human histological sections. Ten healthy subjects were investigated using q-ball DTI data sampling scheme. Shannon's entropy equation was used for entropy calculation using in-house software written in Matlab. In addition, the dependence of entropy with number of diffusion directions was also tested using 15, 25, 55, 70, and 90 directions in diffusion gradients. Entropy revealed enhanced dynamic range of contrast compared with FA. Gray matter regions are more visible in an entropy map than in a FA map Compared with FA, the relative values of entropy with respect to the Corpus Callosum (CC) were approximately triple in gray matter and double in Frontal white matter (FW). The entropy approach also exhibited dependence on axonal density. There is a significant correlation (r = 0.91 and p < 0.04) between entropy values and axonal density measured in different brain structures. Unlike FA, entropy is less affected by axonal orientation and more weighted by axonal density.;Applying entropy to stroke patients and traumatic brain injury (TBI) animals also showed promising results. Entropy could identify white matter reorganization with fiber crossing during stroke and TBI recovery. Although FA shows promise in differentiating white matter (WM) reorganized brain tissue from other damaged tissues, FA does provide erroneous information if crossing fibers are predominant in the white matter reorganized region. Since crossing axons are dominant during the early stage of WM reorganization, our data suggest that entropy method may provide information about the stage of white matter remodeling in the injured brain, with increased entropy alone (without FA elevation) representing an early recovery stage of fiber crossing, while the increased FA identifies more mature linear fibers. Diffusion entropy measurement demonstrates a great promise in evaluating WM reorganization during neurological diseases.