Hello SPM experts!
I am trying to estimate the smoothness of my data so that I can give 3dClustSim an accurate number to calculate the number of voxels required to achieve alphaSim correction for multiple comparisons. I have used some code that was recently posted on this listserv (see below), and some code from Andy’s Brain Blog (http://andysbrainblog.blogspot.com/2012/10/group-level-smoothness-estimation-in-spm.html), and finally a very simple line of code to estimate smoothness (in the code below, I only looked at one person's data at a time):
[FWHM_est] = spm_est_smoothness(sqrt_ResMS.img,mask.img)
With all of these codes I get numbers that are far too high, ranging from 2724 – 35, 951!! My understanding is that the numbers I get should be roughly the size of the smoothing kernel I used during preprocessing.
The images I have used to calculate the smoothness include:
1) the square root of the ResMS file VERUS the pristine ResMS file
2) a mask that has been resliced to 1x1x1 (my voxel size) VERSUS a non-resliced mask with the standard 2x2x2 voxel size that MarsBaR outputs
I have also looked at single subject data, and averages of estimated smoothness across subjects, no matter what, I get these crazy high numbers.
Does anyone know why this might happen?
The code I used from the listserv is below (thank you Katharina Wittfeld & Donald McLaren).
Thanks so much!
Tamara Sussman
mask_path ='D:\F\ETEMP1\data_fmri\CalculatingSmoothness\All_sqrt_Res_MS_files\mask.img';
dir_ResMS= 'D:\F\ETEMP1\data_fmri\CalculatingSmoothness\All_sqrt_Res_MS_files';
% list with residual files
sqrt_ResMS_list = spm_select('list',dir_ResMS,'^sqrt_ResMS.*\.img');
% and mask_path is the complete path to your mask defining the ROI
[FWHM_est] = spm_est_smoothness(sqrt_ResMS_list,mask_path)
|