It looks like you've been trying to import an individual volume parcellation (eg. the aseg+aparc.mgz) as an surface atlas on the template brain. But these tools are not designed to work this way. In your screen capture with the scouts, it's quite obvious that this did not work at all.
I don't think these mixed head models are a good solution for you.
Since you don't seem to be clear on the types of files you are manipulating, I recommend you start with something simpler: either a surface head model (in that case, you already have the FreeSurfer parcellations available in the Scout tab), or a volume head model (in that case, you may import a MGZ file as a volume atlas, but only for the same subject).
Keep your analysis simple. Making your source analysis more complicated will only increase your chances of making mistakes.
If you have MGZ files, I guess you have individual data processed with FreeSurfer. You should import these before processing your data.
And once more, I insist on the importance of starting by the following the introduction tutorials before trying to process personal data.