[next] [prev] [prev-tail] [tail] [up]

1.3 Exploring a DMR project

The DMR project just created may look like this, where the first volume is displayed. In this case, it is the b0 volume. This volume contains 75 slices.

PIC

You may explore the data via Options --> Time Course Movie. By clicking the play buttons >, BV will move through the volumes or diffusion directions of your measurement. The Recalibrate button in the Time Course Movie window automatically adapts brightness to the slice you are currently viewing. This feature is added because the intensities of a b0 images are far higher than those of a DW image.

PIC

Recalibrate

Recalibrate the intensities of the current volume

Preload all

Load all volumes in memory (on slower machines)

Time Point

Enter a value here to go directly to a specific volume

Loop

When checked, BV will loop the volumes

First <-> Last

Switch quickly between the first and the last volume of the data set

The slice before and after recalibrating are shown below:

PIC

PIC

1.3.1 Creation of FA and Mean Diffusivity maps

On the basis of the DMR data, it is possible to directly calculate tensors, FA and Mean Diffusivity maps. For background on tensor estimation, see [?, ?] among others. To recap, Mean Diffusivity is defined as

Tr(D ) M D = (Dxx+ Dyy + Dzz)/3 ≡ -----, 3
(1.1)

and is in theory limited to the interval [0, ∞). Fractional Anisotropy is defined as

∘ 1-∘ (λ-1--λ2)2+-(λ2--λ3)2+-(λ3--λ1)2- FA = 2 ----------∘----------------------. λ21+ λ22 +λ 23
(1.2)

Fully isotropic voxels have FA = 0, while fully anisotropic voxels have FA = 1. This is illustrated in figure 1.1.

PIC

Figure 1.1: From left to right: isotropic tensor, where λ1 = λ2 = λ3 and FA close to 0. Oblate tensor where (λ1 = λ2) >> λ3 and prolate tensor where λ1 >> (λ2, λ3). FA of the oblate and prolate tensors might be similar.

Let’s continue with the FA calculation procedure in BrainVoyager:

  1. Tensor Calculation: Click File --> DMR Properties --> DWI data. A dialog pops up, which shows no gradient information, if you did not specify a gradient file during DMR creation:

    PIC

  2. Click Load GRB and load the appropriate gradient file mgh_dti30.grb (see 1.6.1 for details). Accordingly, the data in the gradient table has changed. To set the right value for b, use the fiels in the top right of the dialog. Set volumes 2-31 to a b-value of 800, and set volume 1 to 0. You can do this by directly editing the table on the left of the dialog.

    PIC PIC

    PIC

  3. Click OK and save the DMR (File > Save).
  4. Now click on the Calculations tab.

    PIC

    You have the option here to mask out the background of the image. Please bear in mind that, by masking via a threshold, you always risk losing voxels in the brain itself. It is safer to use a mask based on the anatomical data, which is discussed in section 1.6.3. For now, because of the nice visualisation, check the mask box and click on Estimate to start the tensor estimation. After the calculations, BV will ask you to save the resulting DDT file, containing the tensor information. Save it as human31dir_dmr.ddt. The ddt calculated from a dmr is different from the ddt calculated from a VDW file.

  5. FA calculation: click the Fractional Anisotropy button, and the FA data will be overlayed onto the DMR. You may explore the resulting map similar to one of an FMR.
  6. Mean Diffusivity: click the Mean Diffusivity button.

PIC
(a) FA map
PIC
(b) Mean Diffusivity map

Figure 1.2: Maps calculated on basis of a DMR

In principle, one could start a complete analysis on these maps. For instance, to create a MD histogram of slice 39 (done in Matlab):

PIC

The disadvantage is that we stay in 2-D space. To do an analyis in 3-D, the DMR project has to be co-registered to a VMR, which is explained in the next sections.

[next] [prev] [prev-tail] [front] [up]