[Brains-users] class image 2 descrete - rule of thumb?

Eugene Zeien eugene-zeien at uiowa.edu
Thu Mar 9 13:34:26 PST 2006


Converting a segmented image to a discrete image is a simple
matter of applying thresholds.  Using the menu item from the
brains2 toolbar applies a talairach box to the basal ganglia
region, where an alternate GM/WM threshold is used.

other/CSF thresh = 10 (I think)
CSF/GM threshold = 70
GM/WM threshold = 190
GM/WM threshold = 197 (within 'basal_box')

I do not see an option for 'b2 convert class-image to discrete'
which would allow one to choose alternate thresholds.


b2 load talairach-box /opt/brains2/bin/talairach/basal_box
Object loaded with filter: brains1
b2 convert talairach-box to mask talbox1 talpar1 expand= no
b2 convert class-image to discrete i1 mask= m1
Diagnostic Message: Thresholds 70 190 197


Usage	b2 convert class-image to discrete  <imageID> <mask= mask>

Description	Converts a continuous tissue classified image into a
        discrete tissue classified image. The resulting image
        contains tissue codes from 0-9 (blood, other, etc),
        10 (CSF), 130 (GM), 250 (WM). The mask defines the
        subcortical region where the alternate threshold
        between GM and WM should be applied. If no mask is
        specified then 190 is used for the entire image.

<imageID>	An identifier for an image

<mask= mask>	The mask used to define the subcortical region.


Christoph Christmann wrote:
> Hi all,
> 
> during the segmentation process BRAINS2 not only provides a _05_segm.img 
> but moreover a discrete_*.img the latter representating a real discrete 
> image with classes for WM (white) and GM (grey). We are not sure how 
> this image was created and how to interpret it.
> 
> The User's Guide states that it is possible to manually transform a 
> segmented image to a discrete image. However, we dont know which 
> parameters were used to do this automatically.
> 
> Moreover we would like to know if anyone uses the discrete images in the 
> manual segmentation process in order to improve the reliability or even 
> the validity. Is there a rule of thumb for the paramters (class image 2 
> descrete)?
> 
> TIA
> 
> christoph
> 
> 
> 
> 
> 




More information about the Brains-users mailing list