I thought that both the epi.mnc (larger bounding box) and the smaller mnc
boudning boxes were based on the mni templates, rathert than the talairach
atlas french woman brain templates?
True
Thanks
Jeff Lorberbaum
On Wed, 21 Jan 2004, Marko Wilke wrote:
> Hi Donghoon,
>
> > Can I ask the reason why the default of the bounding box in the writing
> > option for normalizing is a small box [-78 78; -112 76; -50 85]? If I
> > select it, the dimensions of normalized images is different from those of
> > the template image.
>
> True. The default bounding box is smaller than the one the templates
> use. The default one is close, but not quite like the space defined by
> Talairach and Tournoux (I refer you to Matthew Brett's exhaustive
> website on this (www.mrc-cbu.cam.ac.uk/Imaging/Common/mnispace.shtml)
>
> > At first, I just selected the default but I found the dimension of the
> > result images is different from that of the template image and my ROI
> > masks.
> > Because I had to use ROI masking in the same dimension as the template
> > image, at the second time, I selected the bounding box size of the template
> > for the normalizing.
> >
> > What I alarmed, a different choice for the bounding box option made a small
> > difference in the indivisual results and a little bit big difference in the
> > group results.
>
> Well, if you choose a larger bounding box, you will (after
> normalization) end up with larger brains. Depending on what you want to
> do, it may be worth trying either of the following (or something else I
> have not thought of :)
>
> use the default bounding box and match your masks to your brains,
> instead of matching your brains to your masks. Your masks will likely be
> based on a single brain, so you could "normalize" this one to an average
> of your brains after normalization and apply these parameters to your
> masks. Matching the masks directly is problematic since the gray values
> are very much unlike the target.
>
> Blowing brains up more will artificially increase the total voxel count.
> If you are dealing with structural images, I would suggest using data
> modulated with the Jacobian determinant, which (kind of) "undoes" the
> blowing-up (search the archives for Jacobian and you will find enough
> reading for a couple of long winter evenings).
>
> > I'm not sure which results is right, with the default bounding box size or
> > with the template bounding box size?
>
> Both could be right, and both could be wrong :)
>
> > Every comments will be appreciated!
>
> Not sure about this :)
> Best,
> Marko
> --
> =====================================================================
> Marko Wilke (Dr.med./M.D.)
> [log in to unmask]
>
> Universitäts-Kinderklinik University Children's Hospital
> Abt. III (Neuropädiatrie) Dept. III (Pediatric neurology)
> Hoppe-Seyler-Str. 1, D - 72076 Tübingen
> Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473
> =====================================================================
>
|