[caret-users] Visualizing functional data normalized to SUIT

[Donna Dierker]

Just an update for the list:  We have emailed Joern Diedrichsen about 
the lack of origin information in the SUIT toolkit volumes (all except 
T1.hdr).  I doubt this is a real issue for the functional volume, which 
in JD's case does specify a plausible looking origin.  When you map this 
volume onto the Colin cerebellar surface, the results look reasonable 
(see attached capture).  But given that the link between the origin of 
Colin's cerebellar surface and JD's functional volume is a bit of a 
black box to me, I don't consider it bankable yet.  Joern will probably 
clear this up in time, and then I can draft a mini/rough tutorial for JD.

In this case, I simply opened the SUIT spec; loaded all files; opened 
JD's volume; mapped to Caret (loaded fiducial surface); and selected 
enclosing voxel.  Note that the hottest spots of cerebellar activity are 
deep in the nuclei (i.e., don't intersect surface).

Also, as Joern's page explains, colin's cerebellum is but a single 
subject, and thus is more biased than their multi-subject template.  
Don't be surprised if Colin's cerebellar surface doesn't intersect your 
group results as nicely as if we had a SUIT average cerebellar surface 
(and don't hold your breath on that one).  Still, I think this is kind 
of cool.

On 04/21/2008 09:46 AM, Donna Dierker wrote:
> Hi Janine,
>
> Your message was addressed to caret-users-owner at brainvis.wustl.edu, 
> which is aliased to me.  But in order for it to go to all of the 
> caret-users mailing list, the address should be 
> caret-users at brainvis.wustl.edu, which I have cc'd.  If you are not a 
> member of caret-users, then you probably won't see replies from other 
> users (e.g., Joern Diedrichsen, John Harwell or David Van Essen), 
> unless they explicitly paste your address in the To or CC fields, 
> rather than simply replying all.
>
> Anyway, I have a few inline replies of my own below.
>
> Donna
>
> On 04/21/2008 04:40 AM, J D Bijsterbosch wrote:
>> Dear CARET users,
>>
>> I would like to visualize functional MRI group results to a flattened
>> cerebellar
>> atlas. I have normalized my images using the SUIT cerebellar template 
>> and would
>> like to visualize this using CARET. The SUIT package provides a spec 
>> file and I
>> have attempted to visualize my results using 'map volume(s) to 
>> surface(s)' in
>> CARET. However, my results are not the same as they are in SPM5.
>>   
> I had to google it, but I see this is Joern Diedrichsen's contribution:
>
> http://www.bangor.ac.uk/~pss412/imaging/suit.htm
>
> Looks pretty cool!  I knew about the Caret toolbox, but much of this 
> is news to me.
>
> This section looks expecially relevant:
>
>
>    Spatial unbiasedness, and the relationship to MNI space
>
> I think there's a pretty good chance Joern's still on the mailing 
> list, so I hope he might help.
>> I am unclear about the steps to be taken and the meaning of for 
>> example the
>> thresholding and color-coding in the final image. The tutorials and 
>> quick
>> guides are not very helpful as they only tell you what to click for the
>> standard case with provided data and do not explain the meaning of 
>> each step.
>>   
> Thresholding in Caret is straightforward -- perhaps too 
> straightforward at times.  For example:  Say you are trying to 
> visualize a map of t-stats overlaid on your anatomical volume.  Then 
> entering "+3.0" for the positive threshold in the metric settings page 
> will make any voxels with t-stats less than 3.0 display with no 
> overlay.  Only voxels whose t-stats exceed 3.0 will have overlay 
> coloring.  Caret does not convert stat volumes to p-values, so if you 
> don't see activation where you expect it to be, then look for these 
> common causes:
>
> * The origin or orientation of your functional volume doesn't match 
> the surface/anatomical.
> * You're trying to map a NIfTI .hdr/.img pair, which Caret doesn't yet 
> support.  Convert it to .nii.
> * The palette is scaled such that your values are in the zero bin.
>
> If you toggle on Display Color Bar on the metric settings page, the 
> min and max of the color scale will be shown, along with the whole 
> color spectrum.  There are several built-in palettes available, or you 
> can make your own palette 
> (http://brainmap.wustl.edu/caret/caret_help/file_formats/file_formats.html#paletteFile).  
> The Attributes: Palette: Edit menu gives you a graphical user 
> interface for this purpose.
>
> Have you had a look at this tutorial:
>
> http://brainmap.wustl.edu/caret/pdf/Caret_5.5_UserGuide_Analysis.pdf
>
> It is less heavy on the scenes and more nuts and bolts oriented, I find.
>> I hope someone can help clarify the steps in CARET so that I can 
>> visualize my
>> results successfully.
>>   
> If no one else has more specific suggestions within the next few days, 
> and the tutorial above doesn't address your questions, then feel free 
> to upload one of your group results volumes that you wish to map.  
> I'll give it a shot and let you know how I fare.
>>
>> Best wishes,
>>
>>
>> Janine Bijsterbosch
>>
>>
>> Academic Clinical Psychiatry
>> University of Sheffield
>> United Kingdom
>>   
>
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: cerebellum.jpg
Type: image/jpeg
Size: 164297 bytes
Desc: not available
Url : http://brainvis.wustl.edu/pipermail/caret-users/attachments/20080425/6f4120fd/attachment.jpg