Hi Joao,

previous bias correction should be fine, and previous skull stripping (if it's accurate) should as well, assuming you understand the recon-all stream well enough to get into it at the right place.

As for the false negatives, can you tell us something about your data? What stage of AD? What are you comparing to? We've processed a *ton* of AD data and have seen different things in different cohorts, but some basic patterns repeated.

cheers,
Bruce

On Sun, 5 Oct 2008, Joao Pereira wrote:

Pedro,

Thanks for your reply.

By false negatives I mean a lack of detection of atrophic areas which I know
are present in this AD cohort, namely the parietal convexities. I'm unable
to visualise them with FS.

Finally, you've recommended me not to seed anything to recon-all which had
been skull stripped. Is there any way of going around this? And what about
data which had been previously bias corrected?

Thank you very much for all you help!

Best,

Joao

 -----Original Message-----
 From: freesurfer-bounces@nmr.mgh.harvard.edu
[mailto:freesurfer-bounces@nmr.mgh.harvard.edu]On Behalf Of Pedro Paulo de
Magalhães Oliveira Junior
 Sent: 03 October 2008 21:25
 To: Joao Pereira
 Cc: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] Voxel resolution impact / skull stripping and
biascorrection replacement


 João,


 Here is my opinion:


 I have worked on resolutions distinct than 1mm3 isotropic and usually I
have no problem. What do you mean by false negatives in your cohort study?


 I think you should avoid seeding data with skull striped by other program
to the pipeline.


 Best Regards,


 PPJ



 2008/10/3 Joao Pereira <jmsp2@wbic.cam.ac.uk>

  Hello,

  I have a couple of questions on which I would like to have your opinion.

  The first is the voxel resolution impact on cortical thickness
sensitivity.
  I'm using 1.25 mm3 isotropic voxels, and I know that FS interpolates
scans
  to 1mm - is it possible that the 1.25 mm resolution hinders cortical
  thickness measurements? I'm asking this because I'm analysing an AD
cohort
  whose result seems full of false negatives (I'm in the process of
inspecting
  the intermediate steps as well - I just wanted to check if resolution
was a
  factor as well).

  I would also like to know if it is possible (and advisable) to input
already
  optimally skull stripped and bias corrected brains into the recon-all
  pipeline. If that is possible, I would assume that I need to bypass
those
  steps, and I would like to know which arguments to use in recon-all in
order
  to do so.

  Thank you very much for all your help!

  Best,

  Joao Pereira

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




 --
 -----------------------------------------------------------
 Pedro Paulo de M. Oliveira Junior
 Diretor de Operações
 Netfilter & SpeedComm Telecom