Diffusion tensor imaging is used in magnetic resonance imaging to attempt quantify the diffusional direction of water on a voxel-by-voxel basis. The standard method to do this is to apply diffusion encoding gradient along multiple gradient directions during a spin-echo pulse sequence and then calculate the water diffusion from each voxel based on the acquired set of images.

An image with the primary direction of water diffusion is color encoded. Red is left-right diffusion, green is up and down the image and blue, is in and out of the image.

A map of the fractional anisotropy (FA). Brighter areas correspond to regions of high anisotropy (i.e. they are preferentially oriented) and darker areas correspond to regions of low anisotropy (or high isotropy, which means there is no preferential diffusion direction).

The apparent diffusion coefficient (ADC) which is the average amount of diffusion per unit time.

The way I like to think of all of this is that if you look at a map of Canada’s lakes and rivers, the lakes would be regions of low anisotropy (low FA) and the rivers would be regions of high anisotropy (high FA).

What is interesting is to look at the underlying data acquired from which the FA, ADC and colormap images are calculated. You can see the diffusion encoded data here which was used to calculate the DTI images on this page.

The equation for the b-value is b=γ2 g 2 δ2 [ Δ - δ/3]. For example, if δ=5 ms, Δ = 10 ms, g = 25 G/cm and γ = 2.675×10^8 /T/s then b=931 s/mm2 (with the proper unit conversion..)

Note: You can do this calculation in Google, for example: (2.675*10^8 /(tesla*s))^2 * (25 gauss/cm)^2 * (5 ms)^2 *(10-5/3)ms in s/mm^2