Hi Lars,

the processing in Han is outdated since many years. It has been replaced by the longitudinal stream. It not only registers all time points into a mid-space (subject template space) to avoid introducing processing bias. A bias sneaks in, when you prefer a specific time point, e.g. as the common registration target (see our papers under 2. References here https://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalProcessing ). Also we developed a dedicated robust registration tool to account for longitudinal changes and get much better alignment than with FLIRT. And finally we do not simply map/register surfaces, we create surfaces in each time point starting from a common initialization, to reduce noise.

Best, Martin

On 04/27/2016 11:46 AM, Lars M. Rimol wrote:

Hi,

I wish to test the reliability of series of three T1 scans that were collected over a period of a few months. The Han et al., 2006 paper recommends using FLIRT to do a linear registration between the scans and then the volumetric registration matrix is applied to the corresponding surfaces (please see quote at the bottom of this e-mail).

I was wondering if this is the way to reproduce what was done in that paper:

1. recon-all -all on all three scans
2. register e.g., the 001.mgz from tp2 and tp3 to tp1 using FLIRT;

which timepoint was chosen as the reference tp in Han et al.? Or was some type of average template created? Registering all other images to one timepoint seems contrary to the thinking behind the FreeSurfer longitudinal stream (which of course didn't exist at that time). Would you recommend using the FS longitudinal stream in stead? 3) apply the registration matrix to 'h.area and *h.thickness. Here the question is how do I apply the registration matrix to the area and thickness files? I saw mris_apply_reg was recommended in order to register to xhemi.rh.w-g.pct.mgh but I'm not sure if I can use mris_apply_reg here?

from the 2006 paper: "[..] a simpler linear registration is adopted to perform the intrasubject surface alignment, and surface point correspondence is then built according to their Euclidean distance in the registered space [..]. The linear registration is computed using the FLIRT linear registration tool (Jenkinson et al., 2002). The volumetric registration matrix is then applied to the corresponding surfaces."

Thank you!

sincerely yours,

Lars M. Rimol, PhD Senior researcher, Norwegian Advisory Unit for functional MRI Department of Radiology, St. Olav's University hospital, 7006 Trondheim, Norway

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer