Dear Alle Meije Wink,
Thank you for your answer. Unfortunately, I am afraid the thickness of the
slices doesn't account for the difference in this case: we only partly scanned
the brain, and the slices are 6mm thick, something quite usual for that kind of
data. As to our area of interest within those images, it turns out that they
include the whole eight slices -- so we cannot afford losing information on any
of them.
As Stephen Fromm asked, I have included the two previous *.img, this time with
their corresponding *.hdr. I hope it will help figuring out where the flaw is.
Thank you once again.
Sincerely,
Guillaume
>
>Dear Mr. Marrelec,
>
>> If you visualize the resulting files (two of which I have attached to this
>> mail), you'll clearly notice that there is something wrong. Actually, on
every
>> realigned volume that I have checked, the first and last slices are very
>> strongly artifacted. I went through the process twice, and I got the same,
>> disturbing, result twice, even though the curves included in the "spm99.ps"
file
>> don't show anything abnormal.
>> Can this be related to the fact that we only have 8 slices per volume? We
have
>> never had this problem before, when treating acquistions with 14-slice
volumes.
>
>Isn't this the result of a rotation in the saggital and/or coronal
>plane? If the slices are very thick (which I can imagine if you have 8
>slices/volume) and the image is rotated, it has to use data from two
>planes (which may be very different) at some points, and at other points
>it only uses data from one plane. This could be the cause of the effect
>in your top and bottom slice. It doesn't matter much as long as your
>area of interest is close enough to the centre of rotation.
>The reason you don't see this on your .ps file is that it shows the
>planes that intersect at the centre of the volume (defined in the spm
>record). At this points the 1/2-planes problem practically never occurs.
|