Dear SPM experts,

I have a question regarding to the procedure of small volume correction.

The traditional procedure for SVC may be as follows:

1)       Use no mask in the “specify 2^nd-level”;

2)    estimate the parameters;

3)       choose "none" during "p value adjustment to control", followed by a threshold, either a p value like .001 or a T value;

4)       define the SVC which could be a sphere, a box or an Image (mask)

5)       finally look at the updated results section

Here, I would like to ask that is the following procedure right for performing SVC?

1)       Define the mask from prior knowledge;

2)       Use this mask as an explicit mask in the “specify 2^nd-level”;

3)       Estimate the parameters;

4)       Finally perform multiple comparison correction.

The only difference of the two methods above, in my experience, is that the former estimate the smoothness uses data from the whole brain and the latter estimate the smoothness uses the data in the explicit mask.

The latter is correct or both of them are correct. If only the former is right, what is the underlying rationale of estimating the smoothness using data from the whole brain rather than the region (small volume) I focused on? In my personal sense, the latter looks more reasonable than the former.

Any help will be greatly appreciated!

Best,