Hi Abigail,
If you choose to go with the first method, the resulting brain mask from the first level analysis would depend on several factors:
- whether you included an explicit mask
-the size of your explicit mask
-and ultimately the defaults mask threshold (which would determine which voxels would be considered in the search region). SPM has a defaults mask threshold of .8 (check spm_defaults file), which is quite conservative. So even when you use a bigger explicit mask, SPM will only include voxels that meet the .8 threshold (try running 2 different masks with the same threshold and you will see what I am talking about). It may be beneficial to do some research on what would be a more appropriate threshold for your study. Some programs (such as specmask.m; can be found in the forum somewhere) will give you the option of changing this value to -inf in the SPM.mat file (prior to running model estimation). However, since this means that any voxel with a value greater than zero will be included in the model estimation, you want to make sure that you have a really good mask.
In my experience, it may not be a good idea to use a structural segmentation (this is the c1* segment that you get from the SEGMENTATION preprocessing step) since you would omit important brain voxels in the first level analysis. A better way of addressing this problem would be to specify a brain mask that would encompass both gray + white matter. You can:
1. select SPM's brainmask. Keep in mind that if you wish to use the -inf threshold, you should convert this to a binarised image since the brain mask is smoothed
2. create your own by skull stripping the structural image for each individual and using it as the mask.
3. creating an average image of all your subject's skull stripped structural image.
These were all options that we tried in our lab that resulted in good masks. Hope that helps.
Best,
Vy
________________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] on behalf of Abigail Livny-Ezer [[log in to unmask]]
Sent: Monday, September 27, 2010 1:57 PM
To: [log in to unmask]
Subject: [SPM] brain mask
Hi
I posted this message, a few days ago but received no reply.
I would appreciate any help!
"I am trying to figure out how to use a whole brain mask.
There are two options to add masking: one in the fMRI model specification module, in which it is possible to add an explicit mask and the other is in the results module, in which there is an option to choose an image. I am not sure which option is the correct one if at all.
If I need to choose the fMRI model specification module, the manual offers to select a segmented structural image. Which image is this? Is it the ms_*.img?
If I need to choose the second option in the results module, do I call the mask.img created in the model estimation or should I call the image provided by spm in the apriori folder called brainmask.nii?"
Thanks
Abigail
|