Progress in magnetic resonance diffusion image

Abstract

Diffusion of water is based on the principle that water molecules show random motion. In diffusion studies on magnetic resonance imaging using a motion proving gradient, a signal decrease which is caused by molecular diffusion in the tissue provides information about tissue structure and physiological status. Diffusion-weighted images play a very important role in brain related diagnosis, especially for acute infarctions, and have become an essential tool in clinical practice. Diffusion tensor imaging using the Gaussian distribution model also has made strides in evaluating anisotropic water diffusion in the white matter in which water diffusion is restricted by macromolecules, membrane, and myelin. Diffusion tensor imaging is based on a mathematical model of an ellipsoid. Thus principal directions of diffusion and the magnitude of diffusion along those directions can be calculated. Diffusion tensor tractography is an innovative application of the diffusion tensor method which can access the pathways of neuronal fibers. However, neuronal tissue is a highly heterogeneous structure on various scales and the Gaussian distribution has only been adapted for free diffusion in a fluid that is uniform and sufficiently large. Complicated wall structures exist in the neuronal tissue, and the Gaussian distribution model does not reflect the existence of a large number of compartments and nerve tracts, particularly in the central nervous system. Therefore, instead of a simple Gaussian distribution model, measurement of the probability distribution of a detailed water molecule is necessary. One method is q-space imaging (QSI). QSI is performed by measuring a large amount of diffusion encoding and can provide the probability density function of individual water molecules. Although QSI is theoretically superior to conventional Gaussian distribution analysis, one limitation of QSI is the long acquisition time due to the large sampling number. The concept of kurtosis is another method to evaluate non-Gaussianity. Kurtosis is a quantity that reflects a gap between the diffusion distribution of a water molecule in vivo and the Gaussian distribution, and can be calculated from datasets of relatively few diffusion encoding elements compared to QSI. Thus, a rather short acquisition time is required for kurtosis. Recently, other innovative applications of non-Gaussian diffusion have been introduced including neurite orientation dispersion and density imaging (NODDI) which is a model-based diffusion technique that allows for the quantification of specific microstructural features directly related to neuronal morphology. These novel techniques will provide useful information on tissue microstructure not only in the scientific research but also in the clinical practice.