That looks pretty good to me considering the resolution and 7T. We will be re-evaluating the hires stream over the next year. If you want to contribute a problem data set to our test suite, that would be appreciated.doug
On 4/17/17 1:45 PM, Damien MARIE wrote:
Hi,
I have a project on 7T data. I processed 26 subjects with FreeSurfer6 recon-all with or without the hires flag. The input is MP2RAGE, 0.6 mm^3, bias corrected, skull strip was done despite a slight piece of dura is still present.For the data processed at native resolution, with the hires flag, I mainly used this: recon-all -subject -i -parallel -expert -hiresThe expert file indicating only mris_inflate -n 50
As a result, skull strip had to be fixed for 5 subjects (missing brain pieces). Bilateral temporal pole is missing in the reconstructed surfaces in 19 subjects.For the downsampled data, without the hires flag : recon-all -subject -i -parallel Bilateral temporal pole is missing in the reconstructed surfaces in only 1 subject. Surfaces are way smoother.I guess such differences between the classic pipeline and the pipeline tuned for hires data are expected. I was just wondering if I was missing something somewhere, some additional options that I would need to tweak, to get better surfaces with the hires data especially in the temporal pole. For now the only things I see that I could do are adding checkpoints or play with this command line: mris_make_surfaces -max_gray_at_csf_border X -max_csf YThank you and best,
*Damien*