Thank you for your quick reply! I have tried setting the absolute
threshold as low as 0.01 and even running the analysis without masking.
The results are still offset. The problem is that even if I optimize the
masking I can not use the glass brain to get the right talairach
coordinates for my results, since they are shifted downwards.
Do you or anybody else have an other idea of what to do?
Carl Johan Ekman, MD
On Fri, 11 Apr 2008 00:37:49 +0900, Kiyotaka Nemoto <[log in to unmask]>
>I think your problem comes from setting threshold for masking.
>Please check "mask.img" in your working directory. I think your mask
>image will be similar to the result you got.
>Mask image will be derived from the value you set for threshold masking.
>What value did you set for threshold?
>Dr. Christian Gaser recommend you start with absolute 0.1 and increase
>up to 0.25 to decrease search volume.
>Another way is to make explicit mask for yourself.
>Dr. D. Gitelman shows how to make explicit mask.
>Anyway, please check the mask first, and if it is smaller than it should
>be, try above.
>On Thu, 10 Apr 2008 14:46:53 +0100
>Carl Johan Ekman <[log in to unmask]> wrote:
>> Dear VBM and SPM users!
>> I have run into a severe problem concerning the presentation of
>> VBM results. The results are shifted dowmwards compared to the glass
>> (see file "result.jpg" just a dummy test for illustration of the whole
>> brain). In the mailing list I have read that others have
>> had the same problem but when using their own made templates. I am using
>> the standard templates for VBM5.
>> I have checked that all images are aligned to the AC-PC line before
>> I have checked all smwc1*.img images with Check Reg and compared them to
>> the templates, they all have the same orientation and size.
>> I have run analyses without the implicit mask to make sure that one
>> image does not mask the others.
>> I tried using VBM5.1 for segmenting, using the "center of mass"
>> resulting in a slightly smaller, but still, offset (see
>> file "centerofmass.jpg").
>> Does anybody know what is wrong here?
>> Carl Johan Ekman, MD
>> Karolinska Institute
>> Stockholm, Sweden
>Kiyotaka Nemoto M.D.
>Department of Psychiatry,
>3690-2 Kaiharaduka Ryugasaki,
>Ibraki 301-0856 Japan
>E-mail: [log in to unmask]