> The first option is perfectly fine.
> If you've used the -subsamp2 (rather than -subsamp2offc)
> then the flirt matrix that relates the original and downsampled
> image is just the identity matrix. Hence you can apply whatever
> transformation was true for the original images to any combination
> of the downsampled images and it will be accurate.
OK, thanks for the explanation.
> As for references, I can't really help here as we do not
> normally need to apply any downsampling. Are you simply
> doing it due to lack of memory?
Yes -- its due to lack of RAM.
Thank you very much,
> On 9 Jul 2010, at 19:33, Matthias Ekman wrote:
>> Dear Experts,
>> for computational reasons I had to run some calculations on
>> downsampled functional data. As result I got some zstat images which
>> should be registered to the MNI152 template. The downsampling was
>> achieved using fslmaths -subsamp2.
>> Before I start "playing" with the parameters on my own I want to ask
>> you which way you would suggest for doing this.
>> Currently I am considering 2 possibilities (preferring the first one):
>> 1.) using the existing 'examplefunc2standard.mat' I got from the non
>> downsampled registration
>> 2.) apply the downsampling to my func_data, T1 and MNI template.
>> Register the downsampled func_data to the downsampled T1. The
>> downsampled T1 to the downsampled MNI. Concatenate both matrices.
>> Register the downsampled MNI to the non-downsampled MNI and finally
>> apply this matrix to the downsampled func_data in the downsampled MNI
>> Would be also helpfull if you point me to a reference/article where
>> downsampling + registration was used.
>> BTW -- Is it common to apply spatial filtering to the original data
>> 'before' downsampling?
>> Thanks in advance,
>> Matthias Ekman, PhD Student
>> Max Planck Institute for Neurological Research
>> Gleueler Strasse 50
>> 50931 Cologne, Germany
>> E-mail: [log in to unmask]
>> Phone: +49-(0)221-4726-255
Matthias Ekman, PhD Student
Max Planck Institute for Neurological Research
Gleueler Strasse 50
50931 Cologne, Germany
E-mail: [log in to unmask]