[VoxBo] vbdumpstats and timeseries question

[Ranjani Prabhakaran]

Hi Dan,

I just tried preprocessing this subject's files again and then ran  
the glm again as well.  When I try to load the glm and a mask using  
idl, I'm getting the same error message as before:


>> (21) No time-series available for requested region
>> [called from: READTIMESERIES
>> </usr/local/VoxBo/pros/VoxBo_10.pro(1712)>]
>>


I'm using the same preprocessing script that I used for all of the  
other subjects in this study, and I can't seem to figure out the  
source of this problem.  The mask is in the inferior frontal gyrus,  
which should be well within the area of functional coverage for the  
subject.  I ran vbqa on this subject's tes files and it didn't  
mention anything out of the ordinary.  But you mentioned that the  
pattern of the prm file suggests that at least one of the tes files  
is out of alignment with the rest.  What in the prm file tells you this?

Any ideas as to what is causing this mis-alignment? I've tried pre- 
processing a couple of times and re-running glms and I'm not sure  
what to do next.


The glm lives here: /xjet/prabhakr/michigan_celeb_cfn/s16_cfn/models/ 
s16_collapsed_gs
And a test mask lives here: /xjet/prabhakr/michigan_celeb_cfn/s16_cfn/ 
Anatomy/s16_LIFG_testmask.cub
And the pre-processing script lives here: /xjet/prabhakr/bin/bin_cfn/ 
vbprep/michceleb_vbprep/s16_postinterpolate_michceleb.vbp


Thanks so much for your help!

- Ranjani
On Jun 23, 2009, at 12:23 AM, Daniel Kimberg wrote:


> Ranjani  wrote:
>
>> I'm trying to use vbdumpstats to get ROI beta values.  I saved a
>> functional mask for a subject and then used vbdumpstats to get the
>> beta value for the voxels in that mask.  According to vbdumpstats,
>> there are 48 voxels in that mask.  But when I loaded my .prm file
>> using idl, applied the mask, and checked the number of voxels - there
>> are supposed to be 101 voxels in the mask.
>>
>
> The numbers can disagree if the mask includes voxels that lie  
> outside your functional coverage, as they do in this case.  I took  
> a look in the directories you indicated, and found a mask that  
> contains 101 voxels, but apparently only has a 20 voxel overlap  
> with your functional coverage (as indicated by vbdumpstats, and I  
> double-checked it another way).  I couldn't get the number 48 to  
> pop out, maybe that's from a different mask or different subject?   
> Note that vbdumpstats will include a voxel only if its represented  
> in all of your image data files.
>
> It's possible something went wrong in the preprocessing here,  
> because I took a look at the prm file, and there's a distinct  
> pattern that suggests at least one of your tes files was out of  
> alignment with the rest.
>
>
>> Also, when I use the idl "Find All" tool to find out how many voxels
>> are in the mask, I've been getting the following message:
>>
>> (21) No time-series available for requested region
>> [called from: READTIMESERIES
>> </usr/local/VoxBo/pros/VoxBo_10.pro(1712)>]
>>
>> The glm for the subject ran without any problems and I can load
>> the .prm file in idl, so I'm not sure what the source of this problem
>> is.
>>
>
> Sounds like it's probably just because your mask includes voxels  
> for which at least one of your tes files has no data.
>
> dan
>