yes, using T2 or FLAIR is a better solution, or using multi-echo mprage so you get the late echoes without sacrificing CNR
On Tue, 12 Feb 2013, XJ Kang wrote:
Hi FS Experts,
I post an email one year ago about problem of the segmentation of GM/WM because of the duram layer. (See https://mail.nmr.mgh.harvard.edu/pipermail//freesurfer/2011-February/017251....).
I found in the FreeSurfer release notes that the V5.2.0 will have flags to refine the pial surface using T2/FLAIR images. I will test those when the new version is available.
What I did so far to solve the problem are: -> Increase TR: CSF darker, dura darker, but scan time longer. -> Increase receiver BW to change TE : TE cannot be changed in MPRage. Increase BW will reduce TE, but also reduce SNR of everything, including dura.
As a result, the brain segmentation seems to be good but the scan is longer and SNR is lower. See changes from brain1.jpg to brang2.jpg. brain1.jpg: MPrage, TR=1900ms, TE=2.45ms, Voxel: 1x1x1mm, BW=220Hz/Px brain2.jpg: MPrage, TR=3000ms, TE=1.62ms, Voxel: 1x1x1mm, BW=340Hz/Px
Anybody has experiences or comments on this problem? Hopefully the flags "-T2" and "-T2pial" in the new version are a better solutions.
XJ Kang