Deryk,
You may be limited only to those tracts anyways. I think with 5mm slice
thickness you might have quite a bit of difficulty using TBSS successfully
(for example, I don't know how well the skeletonization preprocessing step
will work with that much partial voluming). I know that a group here at
Emory has been trying to do TBSS on 1.5mm isotropic macaque brains
(anatomically equivalent resolution in humans=3.5mm isotropic) and has been
having a lot of difficulty getting TBSS to skeletonize the smaller tracts,
because of partial voluming. Then there maybe other pitfalls further down
the processing line that I am not thinking of, particularly because the
partial voluming will be directed in only one of the three coordinate axes.
Peace,
Matt.
-----Original Message-----
From: Deryk Beal [mailto:[log in to unmask]]
Sent: Thursday, March 06, 2008 11:32 AM
To: [log in to unmask]; Matt Glasser
Subject: Re: Eddy current correction before or after resampling?
Hi Matt
I believe it was upsampled by the scanner. The message I have from our MRI
physicist is "true res was originally only 128."
This data set was acquired on a GE Sigma Excel 2 years ago - prior to the
Smith (2007) publication of TBSS recommended protocol. I want to
downsample in preparation of the data for TBSS. I understand that it is
preferred to have the voxel size as close to isometric as possible. My
dataset is 256X256X27 with 5mm thick slices. My voxel size is 1X1X5. I was
thinking resampling to 128X128 would move me closer to an isometric voxel
size as well as reduce the CPU processing time for TBSS.
What is more important - isometric voxel size or a voxel size < 3 mm? I
ask with regards to considering resampling to 64X64 for a voxel size of
4X4X5. However, then I would be limited to TBSS of only the thickest
tracts - correct?
Thanks
Deryk
|