Print

Print
On Wed, May 23, 2012 at 6:10 AM, <[log in to unmask]> wrote:

Dear Dr Mclaren,

 

Thank you so much for your kind help. 

 
"You can either save it with the save button from the results screen or you use can my peak_nii toolbox (www.martinos.org/~mclaren/ftp/Utilities_DGM/) to threshold the images."
 
Did you mean that to load the contrast of interest and then click save? In fact, doctor, I wonder whether this can be done automatically via a code or sth cause it is quite tedious to load this per subject.

Yes. That is what I meant. Peak_nii.m (in my toolbox) can do it automatically.
 

 

P.S.

A question just popped up as I was reading your replies. In fact it is unrelated to the current subject but I am just quite curious about it so I will be really grateful if you could provide me with some insight

 

To what I know, conjunction analysis is only possible when a 2nd-level factorial design model (ANOVA, Full-Factorial, Flexible-Factorial) is used in the SPM context, which gives the overlapping brain regions activated in both two contrasts.

 

Therefore, it seems to me that this can be achieved alternatively by inclusively masking the thresholded t-map of one contrast( obtained in the way as described in your last response) over the other.

You could do this, but make sure the mask is applied at the results stage AND you don't use small-volume correction (SVC). If you use SVC then the results of masking A with B will be different than masking B with A.

 

Did I understand the conjunction analysis correctly?


Yes. However, I prefer the logical AND approach. Save the thresholded map from each contrast. Then use imcalc to compute the overlap. I think its easier to compute and it can show you the non-overlapping areas. However, do not interpret the non-overlapping areas as being different between the two contrasts or conditions. They simply are both not different than 0.

 

I wonder whether my previous understanding is too simplistic because on second thought, in the SPM conjunction analysis, we need to set a threshold if I remember correctly. What does this threshold stand for?

There are several conjunctions. The inclusive masking/logical AND requires a threshold. The global or conjunction null do not require a threshold. The threshold is the statistical threshold you would use for a normal analysis.
 

 

Many thanks and best regards,
                  Ce

----- 原始邮件 -----
发件人："MCLAREN, Donald" <[log in to unmask]>
收件人：[log in to unmask]
抄送人：[log in to unmask]
主题：Re: [SPM] 回复：Re: [SPM] voxel sizes and ROI
日期：2012年05月23日 11点24分