Diffusion tensor imaging (DTI) is a widely used imaging method used to approximate the anatomical structure and structural connectivity of the human brain. Diffusion in general describes the process of the displacement of particles due to concentration differences, i.e. the movement of water molecules through random motion from regions of higher concentration to regions of lower concentration. Since the strength of the MR signal is depending on the phase coherence of the underlying spins, and diffusion processes reduce phase coherence, diffusion can be measured as a decrease in MR signal.

In a diffusion-weighted MR sequence, the spins are first dephased by applying a dephasing gradient and then rephased again. Any molecule that has been displaced within this time period will not contribute to the refocused signal, and therefore the amount of displaced molecules is proportional to the overall signal decrease. The resulting images can thus be used to quantify diffusion.

Diffusion in tissue is restricted depending on the type of tissue, e.g. in cerebrospinal fluid diffusion is isotropic, whereas in white matter diffusion is anisotropic. The anisotropy of diffusion in white matter arises from the fact that diffusion is stronger along axons than perpendicular to axons. Therefore the main directions of the diffusion coefficients may be interpreted as the nerve fiber tracts. Anisotropic diffusion can be described by using a 3-by-3 matrix in each voxel of the image representing all diffusion components in a three-dimensional space. The imaging method based on this principle is called Diffusion Tensor Imaging (DTI).

Data assessed by DTI sequences can be used to calculate various parameters characterizing diffusion processes in the brain, such as Fractional Anisotropy (FA), a measure for directedness of diffusion in brain tissue.

DTI has been used to study the differences in the structure of white matter in healthy and diseased brains including diseases like multiple sclerosis, dementia and psychiatric disorders.