We are doing a longitudinal analysis with three time points (MRI1, MRI2, MRI3). All the subjects were processed well except for one. The MRI3.long of that subject has very bad surfaces (see attachment - I am also attaching the log file). The cross, the template and the first two long runs were reconned well, though. We can't figure out why this third run is so bad. 

As a note, we processed all the cross runs on an Ubuntu machine but we ran the longitudinal pipeline on a computer cluster (both with freesurfer v5.3). We ran the long twice for that subject thinking that the pipeline crashed somehow the first time. However, the results were the same. 

Any help would be appreciated.

Thanks!

Mihaela