v4.3.0 has some randomness in its algorithms which account for these

differences. if you want the results to repeat from run to run, add the

-norandomness flag, or use v4.5.0.

On Wed, 2010-03-03 at 23:27 +0800, liyuanyuan60 wrote:

> HI Bruce and freesurfers

> I have a question about the input file of recon-all

> I follow the tutorial, first use unpacksdcmdir to convert the original

> dcm file to a mgz file,then use the recon-all command with the mgz

> file as the input. I also tried to run the recon-all command with the

> original dcm file as the input. And I compare the output *.stats files

> with different inputs, the *.stats files I compared located in

> <subjectname>/stats/

> Then I found a strange thing, the aseg.stats file between the two

> inputs is the same, however, the aparc file is different，and the

> difference in some index is larger than we thought.

> So, I am not sure, which input file is OK for recon-all analysis?

> I have attached the related results.

> The version we use is

> freesurfer-Linux-centos4_x86_64-stable-pub-v4.3.0

> Thanks for your help!

> ps1, we use siemens machine, so I put *.IMA as the dcm input

> ps2, the scanning orient we use is sagittal, however, my partner has

> asked this question and was answered that we need't reorient it.

> Sincerely, Jin

> 2010-03-03

> ______________________________________________________________________

> liyuanyuan60

> _______________________________________________

> Freesurfer mailing list

> Freesurfer@nmr.mgh.harvard.edu

> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer