Dear Freesurfer experts

I have encountered a problem of performing fully automated reconstruction of my T1 image. Attached please find the resulted image after the reconstruction. The white and pial surfaces do not follow the boundaries as specified by aseg.mgz segmentation. If I interpret correctly, the green line should lay on the boundaries between grey and white matters, and the red line should lay on the boundaries between CSF and grey matter. But that is not my case. The green and red lines seem following wrong boundaries, and it affected the later cortical parcellation stored in aparc+aseg.mgz file. How should I correct or make edit for my case? Is there anything going wrong in my procedure?

For you information, I just have one T1 volume for each subject, and the version of my installed freesurfer is 4.3.0
The command-line I used is recon-all -i <directory that has the COR files> -s <subject_name> -sd <subjects_dir> -all

Regards
Angel