aamod_firstlevel_threshold generating incorrect maps #150

Hello all,

Hope everyone is having a good summer.

I recently ran into some trouble with the heatmaps generated by aamod_firstlevel_threshold. There's some code logic that seems strange to me, but I thought I'd check how other people are using the module before changing it.

The difficulty is here:

        if aas_stream_has_contents(aap,subj,'structural') % Structural if available (backward compatibility)
            tmpfile = aas_getfiles_bystream(aap, subj,'structural');
            if size(tmpfile,1) > 1 % in case of norm_write (first: native, second: normalised)
                tmpfile = tmpfile(2,:); 
            end
        else  % Template
            aas_log(aap,false,'Structural cannot be loaded! Template will be used...');
            tmpfile = aap.directory_conventions.T1template;
            if tmpfile(1) ~= '/'
                if exist(tmpfile,'file'), tmpfile = which(tmpfile);
                else tmpfile = fullfile(fileparts(which('spm')),tmpfile); end
            end
        end

Now:

1. Our analyses always have a structural. The structual stream is not optional nor renameable in aamod_firstlevel_threshold. Ergo, the module will always find a structural (and it will be in a stream named "structural").

2. Our analyses always map the functional results to MNI space. Ergo, unless the structural is normed, aamod_firstlevel_threshold will display the heatmaps on the individual anatomy and it will be wrong. Indeed, I'm getting weirdness like this:

The difficulty is (as far as I can tell) the only stream that is named "structural" that has normed content comes out of aamod_norm_noss. (This is that weird two-part structural stream nobody seems to like and we may change, but that is a separate issue). The issue presently is if you're not using aamod_norm_noss, aamod_firstlevel_threshold will always generate bad maps (This only affects the slices -- the 3D renders are fine, but that is because it always uses the SPM rendered brains).

It seems to me there needs to be a way to force aamod_firstlevel_threshold to use the SPM T1 template even if a structural is present.

On the other hand, it's hard to believe that nobody noticed this before.

Is there something peculiar about my usage?

Regards,
Mike

PS: I think aamod_secondlevel_threshold is similarly broken.