Anisotropic Anomalous Diffusion assessed in the human brain by scalar invariant indices
A new method to investigate anomalous diffusion in human brain is proposed. The method has been inspired by both the stretched-exponential model proposed by Hall and Barrick (HB) and DTI. Quantities extracted using HB method were able to discriminate different cerebral tissues on the basis of their complexity, expressed by the stretching exponent gamma and of the anisotropy of gamma across different directions. Nevertheless, these quantities were not defined as scalar invariants like mean diffusivity and fractional anisotropy, which are eigenvalues of the diffusion tensor. We hypotesize instead that the signal may be espressed as a simple stretched-exponential only along the principal axes of diffusion, while in a generic direction the signal is modeled as a combination of three different stretched-exponentials. In this way, we derived indices to quantify both the tissue anomalous diffusion and its anisotropy, independently of the reference frame of the experiment. We tested and compare our new method with DTI and HB approaches applying them to 10 healty subjects brain at 3T. Our experimental results show that our parameters are highly correlated to intrinsic local geometry when compared to HB indices. Moreover, they offer a different kind of contrast when compared to DTI outputs. Specifically, our indices show a higher capability to discriminate among different areas of the corpus callosum, which are known to be associated to different axonal densities.