Hi there,
I'm new to this software, I have a few questions about using the rest_alphasim GUI.
I suspect most of my queries have been answered before. Where might I find more detailed guidelines on the information the GUI requires?
I read the info posted on the AFNI web page but needed a bit more detail.
This is a list of my questions:
1.       The notes here ( suggest that the cluster connection radius should be equivalent to the voxel size. Is this correct? Which radius should I choose if my voxels are rectangular - I’m using 2.1*2.7 *8 mm voxels, should I use 2mm?
2.       The same notes suggest that if I used smoothing I should use the smoothing kernel in FWHM, is that correct?
3.       I’m not clear on which files to use to estimate the FWHM. I’m working with EEG data and my options are: images from a subject; beta-maps from the group GLM analysis; resMS images from the group analysis. Would they all give me the same results?
4.       I tried to estimate the FWHM and kept getting error messages, e,g.
>> rest_AlphaSim_gui
??? Undefined function or variable "testFlag".
Error in ==> rest_Smoothest_gui>Comput_pushbutton_Callback at 207
if strcmp(testFlag,'Z')
Error in ==> gui_mainfcn at 96
Error in ==> rest_Smoothest_gui at 42
    gui_mainfcn(gui_State, varargin{:});
Error in ==>
??? Error using ==> waitfor
Error while evaluating uicontrol Callback
 Thank you very much! Deborah

Hi Deborah,

1. You want to use such an anisotropic voxel size (2.1*2.7 *8 mm)? I am not very sure if the smoothness could be correctly estimated based on such an anisotropic voxel size. If you use 2mm, then that means face connection (i.e., each voxel has 6 neighbors). I think this way might be acceptable for 2.1*2.7*8 mm voxels.

2. Generally, seems the smoothness should be estimated from the 4D residual files (effective smoothness). From that post (my understanding), if the applied smoothness (FWHM used in spatial smooth) is bigger, then the applied smoothness should be used.

3. From AFNI's post, the best file should be the 4D residual file (which spm usually doesn't output). Another way may be use the statistical image (e.g., T image) as suggested in FSL's post.

4. This follows the FSL's easythresh tradition, a statistical image (e.g., T image or F image generated by SPM or REST) could used as input.