[caret-users] Extent threshold

[Donna Dierker]

There are several nonparametric tests under Caret's Attributes: Metric 
and Surface Shape Statistical Operations menu that are based on Nichols 
& Holmes' Primer Paper models.  Most options have equivalent features in 
the caret_command utility:

   METRIC/SHAPE STATISTICS COORDINATE DIFFERENCE
   METRIC/SHAPE STATISTICS ONE-WAY ANOVA
   METRIC/SHAPE STATISTICS INTERHEMISPHERIC CLUSTERS
   METRIC/SHAPE STATISTICS KRUSKAL-WALLIS
   METRIC/SHAPE STATISTICS ONE SAMPLE T-TEST
   METRIC/SHAPE STATISTICS PAIRED T-TEST
   METRIC/SHAPE STATISTICS TWO SAMPLE T-TEST and WILCOXON

You can specify the number of threads you want to use, if you have a 
multi-processor system.

See the attached caret_command full help for more details, as well as 
the Williams paper:

http://brainvis.wustl.edu/resources/WmsJNsci.pdf
Van Essen, D.C., Dierker, D., Snyder, A., Raichle, M.E., Reiss, A., and 
Korenberg, J. (2006) Symmetry of cortical folding abnormalities in 
Williams syndrome revealed by surface-based analyses. J. Neuroscience 
26:5470-5483.

In most cases, group membership labels are permuted to establish the 
significance area cut-off; for the one sample and paired t-tests, the 
subjects' metric/shape columns are sign-flipped, as described in Nichols 
& Holmes' Primer Paper.

On 08/07/2007 10:43 AM, Anthony Steven Dick wrote:
> Hi Kevin-
>
> We use permutation methods outlined in some recent papers from Nichols 
> and Hayasaka (e.g., see Nichols & Holmes, 2001; HBM 15, 1-25) to 
> determine minimum cluster sizes.
>
> As Donna suggested, the AFNI/SUMA SurfClust command is ideal for this. 
> We use it in combination with R statistical package, and it gives 
> quite good results. The only downside is it takes some time to run the 
> permutations unless you have access to a grid computing system.
>
> Anthony
>
> Donna Dierker wrote:
>> Hi Kevin,
>>
>> Not directly, as far as I know. The closest two such features are these:
>>
>> * When you multi-fiducial map a fMRI volume onto the PALS_B12 atlas, 
>> you can specify a threshold. Then it maps the volume to each of the 
>> 12 PALS_B12 hemispheres, and calculates the suprathreshold area of 
>> each hemisphere. Then it computes an average area and generates a 
>> threshold that limits the MFM extent to that area.
>>
>> * After mapping your functional volume onto the surface and saving as 
>> a metric file (node:scalar pair), then you can open the metric file 
>> as a surface shape file and use Surface: ROI to do a cluster report 
>> on the metric file. I haven't tried it using 0.0 as a threshold, but 
>> it's probably possible to do so. Caret will then find all the 
>> clusters over that threshold. Save the output of that file; trim the 
>> header info; and sort on column 5 (corrected size). (If you're using 
>> a single individual's surface, rather than an average surface, then 
>> cols 4 and 5 will be identical, and you won't need a distortion 
>> metric file; for average surfaces, you need a metric like our PALS 
>> distortion metrics that pump up the area in regions of high 
>> variability (e.g., IPS), that get smoothed out during averaging.)
>>
>> I'm not aware of any more direct methods; you might try the AFNI 
>> list; Ziad Saad may have something like that for SUMA.
>>
>> Donna
>>
>> On 08/07/2007 09:22 AM, Kevin Guise wrote:
>>> Greetings
>>>
>>> I've searched through previous postings and the caret documentation 
>>> but I am
>>> unable to find any information regarding a method to set a minimum 
>>> cluster
>>> extent threshold while projecting functional activation onto a 
>>> surface map.
>>> Is this at all implemented in caret, or can anyone suggest an 
>>> equivalent
>>> method?
>>>
>>> Best
>>> Kevin
>>>
>>> _______________________________________________
>>> caret-users mailing list
>>> caret-users at brainvis.wustl.edu
>>> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>>>
>>>   
>>
>> _______________________________________________
>> caret-users mailing list
>> caret-users at brainvis.wustl.edu
>> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: caret_command.txt
Type: application/x-research-info-systems
Size: 74407 bytes
Desc: not available
Url : http://brainvis.wustl.edu/pipermail/caret-users/attachments/20070807/4f7647f0/attachment.bin