Kev,
I can't speak so much to the method you describe in your e-mail, but for my work I've taken a different approach, owing to certain structures not currently being parcellated in Freesurfer. Of course, in this case I think that's because the DLPFC isn't really well defined, anatomically, as much as it is functionally. Perhaps a better approach would be to frame how you define your DLPFC region of interest based on your hypothesis.
If you do so you'll find that you can either use prior literature coordinates or, and this is what I do, you can use functionally-defined ROIs such as those in the Yeo-parcellations that are now part of Freesurfer - at least for fsaverage. If you look at the 17Network model you can see a few different "DLPFC" ROIs that are based on resting-state subnetworks. You can pick the one you like the most and use various Freesurfer commands to get them mapped to individual subject space or your standard space of choice.
I'd be happy to share my scripts with you if you'd like them.
Cheers,
Paul