Hi Tudor,
You'd add the lgi only to the third step (the -long runs). And you could do a two stage model with it.
Martin

Sent via my smartphone, please excuse brevity.

-------- Original message --------
From: Tudor Popescu
Date:06/10/2014 11:51 PM (GMT+01:00)
To: free surfer
Subject: [Freesurfer] Computing LGI values for the longitudinal stream

Dear FS list,

I did all 3 steps of the longitudinal stream (cross, base, long), plus the long_mris_slopes for my three-group & two-timepoint data set. This gave me files for thickness and area, but I'd also like to do analysis for lGI. To that end, I first did
   recon-all -s subj -localGI
and then
   recon-all -base subj -tp subj_prescan  -tp subj_postscan -localGI
These commands ended with the "abberantly high lGI value" error for 3 subjects out of the total of 72, and for those I then ran what I knew was the standard workaround:
   recon-all -autorecon2-wm -randomness -autorecon3 -hemi ?h -s subj
However received the error
   It appears that this subject ID is an existing base/template from longitudinal processing (-base). Please make sure you pass all necessary flags.

I was therefore wondering if it is in fact the case that the longitudinal stream for lGI is done by simply adding -localGI to the normal commands shown in the tutorial (i.e. cross, base, long + the long_mris_slopes command)?

Also, assuming the lGI values will be eventually be computed as part of the longitudinal stream, are lGI analyses done in the same way as for thickness, i.e. two-stage model and QDEC?

Many thanks for any help.

Tudor