How to quantify results for FSL FIRST, after stats have been generated.
Clusters
For a stats file "L_Hipp_vertexMNI_rand_tfce_p_tstat1.nii.gz" for example, you can get information based on standard FSL reporting (this page has several examples).
The following two commands will get you started:
Defining clusters (see FSL guide)
Example command for corrected p value < 0.05 (which requires a threshold of 0.95 since p values are stored in nifti files as 1-p):
[datauser@localhost temp]$ cluster -i L_Hipp_vertexMNI_rand_tfce_corrp_tstat1.nii.gz -t 0.95 -o cluster_index --osize=cluster_size {optional: > cluster_info.txt}
This will create two nifti files as shown below with values at any location above the threshold (0.95 in the example), one with the cluster index starting at 1 (cluster_index.nii,gz) and one with the cluster size in voxels (cluster_size.nii.gz). The list of clusters will be displayed in the command window, unless the option "pipe" (>) to text file is used, in which case the text file will contain the table information.
Suggestion: run this first without saving to the text file, to verify results, then simply run it a second time with the "> cluster_info.txt" on the end.
Later statistics and calculations refer to the cluster index, so this information is typically used repeatedly.
Tip: if you copy the table and paste into Excel, either from the Linux terminal or from the text file, it will sort into columns properly:
P values and t statistics
Example command to extract t values at significant clusters:
[datauser@localhost temp]$ cluster -i L_Hipp_vertexMNI_rand_tfce_corrpp_tstat1.nii.gz -t 0.95 -c L_Hipp_vertexMNI_rand_tstat1.nii.gz scalarname="1-p" > cluster_p1.txt {optional: > cluster_corrp.txt}
This produces a similar table to the previous example, but with extra columns showing the cluster values extracted from a second file listed after the "-c" flag. The second input file (t statistic in this example) is referred to as "COPE" (see cluster guide).