Hello,
glad one of my suggestions worked for you. However, I do not think you
can easily dismiss those brain regions excluded by the higher threshold
as "certainly not being activated". I would recommend you generate an
appropriate mask including all your voxels and redo the analyses. The
fact alone that the search volume is different may influence your
pattern of activation (smoothness estimation, number of multiple
comparisons to correct for etc.), I therefore don't think you should go
ahead with what you have.
Cheers,
Marko
monika grewal wrote:
> Thanks Marko.
>
> I had tried looking into my raw data and that seemed to be perfectly
> fine. Now, when i changed threshold masking value in spm_defaults.m from
> 80% to 70%, the new mask.img is decent enough.
> That means it was due to this threshold masking only. But, i still want
> to clear if i am right in saying that these missing voxels in mask.img
> actually had intensity values less than 80% of global signal. Hence,
> there's no chance of them being activated. Therefore, even if i change
> threshold masking default value, my results won't be affected much
> (considering that i'm not much interested in looking for deactivations).
> So, i can easily interpret my results based on present analysis
> threshold only.
>
> Or else, should i lower down my threshold masking value and look for
> deactivations in this right hemispheric region.
>
> Regards,
> M. Grewal
>
>
> On Tue, May 7, 2013 at 12:58 PM, Marko Wilke
> <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
> Hello M.,
>
> as you discovered, the generation of the mask relies on all voxels
> being interpretable in all volumes. Hence, if your final mask is
> missing voxels, spm for some reasons thinks it should not use those
> voxels, either because they are missing or because they are not
> interpretable in at least one image. Therefore, the first thing to
> do (which you probably did) is to check the raw data very closely.
> If you cannot find the offending volume, you can play with the
> masking options on the second level by including an implicit mask,
> and/or by disabling threshold masking and implicit masking. Not sure
> that playing with defaults (mask thresh or so) will help, but maybe
> worth a try. After you find the reason, you will then have to think
> again how appropriate your approach was :)
>
> Cheers,
> Marko
>
>
>
>
> M. grewal wrote:
>
> Hello SPMers,
>
> In my block design task, i was getting clear left
> lateralization. While scrolling through individual subject data,
> i noticed that it is not actually so, but merely due to a major
> part of right hemisphere being chopped out in final spmT images.
> Also, i noticed that it is because beta images have NaN values
> at those voxels in right hemisphere (one jpeg image attached
> hereby).
>
> Well, i understand that it is because of implicit masking which
> makes a particular voxel in all volumes either 0 or NaN (based
> on data representation scheme) if its value can't be sampled in
> one volume. No issues if its just one or two voxels. My concern
> here is a big chunk of voxels specifically in right hemisphere.
> Also, this empty voxel effect is consistent over 12 subjects.
> So i'm concerning if its some sort of acquisition issue or
> realignment algorithm issue (well, i'm following standard SPM8
> protocol with most of the defaults intact).
> Also, i have removed implicit masking at realignment and
> smoothing stage. But, i feel this is somehow introduced at model
> estimation level while generating mask.img always. I want to
> know if i could control how mask.img is being calculated and
> modify any thresholding value to include more voxels in
> analysis. Also, i'm not sure how much appropriate that procedure
> would be.
>
> Please help..
>
>
> --
> ______________________________________________________
> 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 <tel:%2B49%207071%2029-83416>
> Fax +49 7071 29-5473 <tel:%2B49%207071%2029-5473>
> [log in to unmask]
> <mailto:[log in to unmask]>
>
> http://www.medizin.uni-__tuebingen.de/kinder/epn/
> <http://www.medizin.uni-tuebingen.de/kinder/epn/>
> ______________________________________________________
>
>
>
>
> --
>
> Monika Grewal
>
> R&D Engineer,
> National Brain Research Center,
> Gurgaon.
--
____________________________________________________
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/
____________________________________________________
|