Hi,
On 27 Feb 2007, at 11:38, Siewmin Gan wrote:
> Hi all,
> I would like to ask your opinion about segmentation of DWI using FSL.
>
> I would like to run segmentation on descalped 3D or 4D DWI images
> and adding
> up the white and gray matter segmented binary images in order to
> remove the
> white rim of CSF in the FA maps which i subsequently derived from the
> processed DWI images.
> My roi analysis are going to be performed on the coregistered and
> realigned
> DWI which are in FSPGR (structural) space.Hence in my study - would
> it be
> better running segmentation of the raw FSPGR and raw DWI
> independently or
> coregister and realign the DWI into FSPGR space 1st and then feed
> the raw
> FSPGR and the realigned & registered DWI into multisegmentation?
> Because my
> DWI slices are thick, the segmentation of the raw DWI is performed
> in 2D.
> Hence, would running multisegmentation yield better results
> compared to
> segmenting DWI images independently?
Given the quality of your DWI images I would definitely guess that
you're best off running the segmentation on _only_ the T1. You could,
however, just try both to see......
> Secondly, to get a brain with white and gray matter and no csf, I
> am only
> able do so by add up the gray and white segmented binary volume
> maps derived
> by segmenting the raw DWI image, using avwmaths addition function.
> When I
> try adding up Partial Volume gray and white matter maps, i get a
> brain image
> which looks wierd ( see attached file). Using AVWmaths, I set a
> threshold
> of 0.9 to each of the gray and white matter partial volume maps so
> voxels
> that are >90% probability of being gray matter ( or white matter) are
> calculated. I then binarise these images and add them up. What have
> i done
> wrongly and any suggestions how I can get whole brain images from
> adding
> partial volume white and gray matter maps?
This sounds like it should be fine. There were no attachments to the
email so I can't see what you sent, however if your DWI segmentation
looks ok (does it?) then combining the hard or soft grey and white
maps should give you what you want.
Steve
>
> Many thanks
> Siewmin Gan
> Department Radiology
> Royal Melbourne Hospital
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|