Configuring the brain extraction parameters

• Main Menu: Modules ->Brain extraction

When you select Brain extraction on the menu line the Brain Extraction Configuration window appears.

Configuring brain extraction parameters.

Parameter configurations for brain extraction.

As a rule of thumb, you should first try to run brain extraction with the default parameters. For most image series the default parameters lead to a good fit of the brain surface.

In some cases, it may be necessary to modify certain parameters. The first group of entries is called Matter parameters. This group contains tuneable parameters for the selection brain and non-brain tissue. Ranges of allowed parameters are given as (low - up) .

Cumulative cuts:

There is an upper and a lower cumulative cut, t_low (0 - t_up) [%] and t_up (t_low - 100) [%] respectively. The unit for these parameters is %. These cuts are used in order to remove a small number of voxels whose intensity values are very different from the rest of the image series. I.e. a fraction of size t_low in the low end of the intensity spectrum and a similar fraction of size t_up in the upper end of the spectrum are not considered in the fitting procedure.

Brain/non-brain matter separation (0 - 100) [%]:

In many MR image bone tissue appears dark. This value is chosen as a rough initial separation between brain tissue (above) and "background" (below). The value represents a fraction of the range between t_low (0 - t_up) and t_up (t_low - 100) and is given in units of %.

Smoothness parameters:

There are two parameters that determine the smoothing of the mesh surface, r_min (0 - r_max) [mm] and r_max (r_min - 20) [mm] . The unit for these parameters is mm. Structures with radii of curvature < r_min are heavily smoothed while structures with radii > r_max experience little smoothing. The amount of smoothing follows a sigmoid function of radius of curvature.

Fractional intensity threshold (0 - 100) [%]:

This parameter represents a fraction of the intensity threshold that separates brain tissue from background in the later stages of the fitting. The iterative update of the tessellated mesh along directions given by surface normals is guided by this parameter.

Threshold gradient (-1 - 1):

The amount of iterative update of the tessellated mesh along directions given by surface normals is guided by this parameter.

Set Centre of Gravity

By default, the centre of gravity used to initiate the iterative search of the image volume is automatically estimated from the input data. If the image volume includes large non-brain regions (e.g. neck), the iteration may not easily converge to a correct extraction of brain. This option enables manual selection of centre of gravity region as well as the initial radius of the search region (see figure below)

Figure 1. Setting size and position of initial search area (‘centre of gravity’ option)

There are two parameters that control the iterations of the brain extraction algorithm.

Maximum number of iterations:

The maximum number of steps used to fit the brain volume.

Maximum number of intersections:

The maximum number of surface self-intersections allowed before refitting the brain surface.  

The Help button leads you to this user’s manual. Upon choosing a set of parameters you may return to the default values by pressing the Reset button. You may close the brain extraction configuration by pressing Close. When you want to start extracting the brain you press the Process button.