In brief: CSPM creates a subfolder "significant_cope" with results files.
Example: V:\LinuxShare\FSL_FIRST\ID111N1045G2\L_Hipp\MNI\TIVagesex\vox\significant_cope
The "cope_tstat..." files are the effect sizes in regions of significant change, in units of voxel size (so 1 mm voxel size means the values are in mm). Bi-directional changes (F-test) are in "cope_tstat1_p05", and "_dec" and "_inc" are decreasing and increasing values separated out, if you want to play both blobs on one surface. The corrp are corrected p value files.
Quick look
Open a p-value file in mricron, and look at the histogram (Ctrl^H); this will quickly show you whether there are significant p values. Remember that FSL saves these as 1-p value, so p = 0.05 will be recorded as 0.95 in the file.
For multiple t or F contrasts, you will need to know what "tstat1, tsta2, tstat3,...etc means; you can open the text design files to verify.
Opening the mask file in a separate, yoked mricron will allow you to ensure you have the cursor on a surface voxel.
The mask file is in the structure/analysis type folder (MNI, native).
Example: V:\LinuxShare\FSL_FIRST\ID111N1045G2\L_Hipp\MNI\TIVagesex\L_Hipp_vertexMNI_mask.nii (GZ file)
Look at values
Look at the min, max, range using the iamge info tool. This information will be useful for putting in ranges for mricros.
Select the effect size map for significant voxels (example below), and you will see the effect size range (max = 0.55 in example below), and how many significant voxels there are (look for non-zero voxels, so 125 in the example below).
Example stat files and design matrix
Consider the following design matrix (design.png), created with Glm:
The image part is the design matrix and the bottom matrix represents the statistics. There are four independent variables in the design matrix, group, Age, Sex, TIV. These are represented by the columns in the design matrix, with the labels at the bottom.
In the matrix, there are four rows, C1, C2, C3, C4. Each row is a contrast, that is a combination of the variables. In this example, the contrasts are all very simple as they are just "1" for one variable and 0 for the rest. For example, C1 is 1 for group and 0 for Age, Sex and TIV. Therefore C1 is simply group.
Note that for FIRST, group should be a single column variable, not two separate columns (not sure why but that's what the documentation says).
One the right of the matrix is a grid of F contrasts F1 - F5. The F contrasts are a combination of t contrasts. This is different to SPM where F contrasts are combinations of variables. For example, F1 is C1, F2 is C2, and F5 is C1 + C2 + C3 + C4. F5 is what SPM ussed to call "Effects of Interest" or "All effects."
The first-pass FIRST statistical analysis is an F test; this give significance maps with corrected and uncorrected p values for each F contrast, and optionally t contrast. In CSPM, the files will be labelled with "N5000" or similar to indicate how many permutations were done for the correction.
However, to get the effect size (betas, or parameter estimates in FSL terminology), we need to run the statistics with different options ("-glm). Since we have already done the correction for multiple comparisons, this has one "N1" for one permutation. We don't use p values from this step, only the "pe" files.
Sorting out the files
The output of randomise are statistic, p value, and corrected p value files, for each contrast. The contrasts to which statistics files correspond are indicated by numbers, e.g., *_tstat1 (or _fstat1), *_tstat2, *_tstat3, and so on. For a two-group design matrix, you will typically have four contrasts. (You create these in the Glm step.) The t contrasts these are listed in the *.con text file created when you create the design matrix with the Glm GUI, and the F contrasts are in the *.fsf files. Below is an example of a *.con file:
CSPM output
CSPM automatically tests for significance, and save files in a "significant" subfolder. The effect size files at labelled "pe" for parameter estimates, and "cope" are corrected parameter estimates.