Display of tensor data by voxel

Tensor values where obtained by an algorithm which uses all available gradient direction in order to reconstruct the symmetric second order tensor which is a model for the water diffusivity in the brain. Using more gradient directions for the computation of the tensor increases the angular resolution of the tensor and counteracts noise in the measurements. Nevertheless the degrees of freedom in a symmetric tensor are reduced from 32 measurements to 6. We obtain in this work more information by validating the tensor representation obtained from a least-mean-squares fit against the raw gradient directions. The model of the second order tensor may not always sufficient to explain the local diffusion thus statistical validation of the gradient directions against the tensor model will indicate regions for which the model fits well against regions where the model fit to the data is poor. Especially the regions where the fit is poor are of interest because they may contain for example crossing fiber bundles. Visualization of the statistical fit is important to guide the interpretation of the data.

Live demonstration

Compute the ADC from the raw data (webpage contains Quicktime movie)

Movies

one slice (webpage contains Quicktime movie)
Cylinders (mpg 35MB)
Ellipse (mpg 27MB)
Lines (mpg 28MB)

Images

Displaying tensor values by cylinders:

301 - tensor display   301 - tensor display

Displaying tensor values by ellipses:

301 - tensor display
Displaying tensor values by orthogonal lines:

301 - tensor display

Apparent diffusion coefficient in false color (301):

301 - apparent diffusion coefficient