Hi David,
I think the reviewer is correct in assuming that the data should not be
resliced more than absolutely necessary. In the case of human data, the
answer would be easier, as affine scaling to MNI space alone will blow
the brain up by a factor of about 1.4 anyway (see Ashburner & Friston,
1997). Interpolation, therefore, is pretty much inevitable at this
stage, and most researchers will take the opportunity of making their
voxels isotropic in the process. For human data, and for analyses in
normalized space (which is only one option), the reviewer's point would
therefore be easier to address.
In your case, however, there are two things to consider: one, I have no
idea of whether the mouse brain template you use is bigger than the
input image, and two, the difference between your native space and your
normalized voxel sizes is a factor of 2 in-plane, but a factor of 4
through-plane. In other words, for every slice in the original image,
you "generate" 4 normalized slices, i.e., you are not only
interpolating, but you are upsampling. You are also massively increasing
the number of multiple comparisons you have to correct for (compared to
a normalized voxel size of, say, .4 mm^3, you have 8 times more voxels).
Whether either of this actually is a factual problem or not, I cannot
tell, but in theory, it could be.
Not sure I have given you a clear-enough answer but I still hope it helps :)
Cheers,
Marko
David Vállez Garcia wrote:
> Dear SPM users,
>
> During the revision process of one of our manuscripts, a disagreement
> between the reviewer and us arise, and I would like to ask for your help
> in this aspect.
>
> To put you in context: we work with animal PET data, which after
> reconstruction have a voxel size of 0.4x0.4x0.8 mm. The scan is manually
> (or automatically) registered to a reference template, and after that it
> is resliced to an isotropic voxel size, in this case 0.2x0.2x0.2mm
>
> I always assumed that after a spatial transformation of the image it was
> advisable to reslice the data to an isotropic voxel, but the reviewer
> consider that /"it is common practice not to change the resolution
> before the analysis is finished"/.
>
> I am trying to obtain some references and good arguments supporting one
> position or the other one. There is someone who can give us some deeper
> insight in the advantages / disadvantages of reslicing the data to
> isotropic voxels after manipulation of the image? I would highly
> appreciate if you can gave us some reference to some paper for further
> reading.
>
> Best regards,
>
> /David Vállez Garcia/
>
> /University Medical Center Groningen/
>
> /Nuclear Medicine & Molecular Imaging/
>
> /Groningen/
>
--
____________________________________________________
PD Dr. med. Marko Wilke
Facharzt für Kinder- und Jugendmedizin
Leiter, Experimentelle Pädiatrische Neurobildgebung
Universitäts-Kinderklinik
Abt. III (Neuropädiatrie)
Marko Wilke, MD, PhD
Pediatrician
Head, Experimental Pediatric Neuroimaging
University Children's Hospital
Dept. III (Pediatric Neurology)
Hoppe-Seyler-Str. 1
D - 72076 Tübingen, Germany
Tel. +49 7071 29-83416
Fax +49 7071 29-5473
[log in to unmask]
http://www.medizin.uni-tuebingen.de/kinder/epn/
____________________________________________________
|