Jennifer,

I think what you are seeing is due to randomness in some of the freesurfer algorithms, as dicom files are converted to mgz during the -i step (which need only be included once) and should itself not account for differences. You can include the flag -norandomness to recon-all which fixes the RNG to one seed. Results should be identical.

In your results, a 1% difference is about right for variability (less than this for thickness differences, I think).

Nick

On Mon, 2008-02-11 at 08:33 -0500, Jennifer C. Britton wrote:

On the freesurfer wiki, it says that recon-all can be done with DICOM inputs in addition to mgz inputs.

In doing recon-all using DICOM or mgz (created via unpacksdcmdir), the volumetric values obtained using mris_volume and mris_wm_volume are different.

For example in one subject, with DICOM inputs: pial 505865 and white matter volume 166713. with mgz inputs: pial 510322 and white matter volume 167030.

The white matter seems to be fairly close but the pial volume is off by nearly 4000.

Can someone explain why there is this difference and which value we should believe?

Thanks, Jennifer

---

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