Many people have asked me about how to obtain reference network files. That is a somewhat personal choice. I used the cortical networks from Yeo et al. 2011, so I will describe how I got those into a format the fslcc command likes. These steps require a functioning freesurfer and FSL installation. I should note that this process was based on input Thomas Yeo provided to our lab on the freesurfer list:
The first thing I did was download the networks provided by Yeo et al. from the following website (download the nonlinear MNI152 space package):
The goal of the next steps is to manipulate the provided images. The first script converts the format of the provided 3d file from freesurfer to fsl 2mm space. The second script extracts each network from the provided file, does a temporal concatenation so each network looks like a seperate timepoint in a 4d file (fslcc needs that), and converts the resulting 4d image to 4mm space (the same as default Melodic output). I suppose it might be better to do the conversion to 4mm space earlier, binarize the 4mm networks, and then do the temporal concatenation, but I don't here.
As with all code, you should carefully audit this before running it on your machine. I’ve tested this process on mac and linux with recent (as of Feb. 2016) installs of FSL and freesurfer.
Transform the Yeo networks from freesurfer space to 2mm FSL space:
Make each network a seperate image, combine networks into 4d file, convert to 4mm space:
Now you can compare melodic output to the Yeo networks.