Dear all,
an additional question on the values that are given in fslstats when using -V to get a volume of a defined ROI...
for example I have a segmented ROI on a DWI pic of a stroke patient and fslstats' ouput is
"163 2865.234375"
Okay that means that I have 163 pixels in that ROI, however what is the unit of the second number (can't be milliliters, that would have to be a whale or sth. like that...)
Greets,
Matthias
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]]" im Auftrag von "Mark Jenkinson [[log in to unmask]]
Gesendet: Mittwoch, 4. Juni 2014 00:08
An: [log in to unmask]
Betreff: Re: [FSL] DWI stroke lesion segmentation in a predefined brain area using a mask
Hi,
It is very difficult to know whether this approach using FAST will work generally on your acquisitions. I would try it and see how it performs. Since acquisitions vary a lot it is impossible to determine whether such an approach will generalise well or not, but my experiences are that such things often need a lot of tuning for different sequences.
All the best,
Mark
On 1 Jun 2014, at 19:03, "Watson, Christopher" <[log in to unmask]> wrote:
> You can manually define the ROI using fslview. Since (I think) you're just going to specify a 3D box, this should be very simple.
> For the FAST stuff, I am unsure, so you will have to wait for one of the FSL developers to give their suggestions.
>
> Are all of your subjects going to be in the acute stage? As you know, the signal abnormality on diffusion imaging will change in a matter of days (perhaps more quickly), so you may get inconsistent results (in terms of, e.g. delineating the "true" infarct core vs. core + penumbra) if different subjects have MRI at varying times since stroke.
>
> ________________________________________
> From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Gawlitza, Matthias [[log in to unmask]]
> Sent: Sunday, June 01, 2014 11:11 AM
> To: [log in to unmask]
> Subject: [FSL] DWI stroke lesion segmentation in a predefined brain area using a mask
>
> Hello everybody,
>
> what I am trying to do is to semiautomatically segment stroke lesions. Basically I would like to do it like Petkova et al. in 2010 (--> http://www.ncbi.nlm.nih.gov/pubmed/21045177).
>
> Besides providing an image (see below) they described their way of segmentation (after BET and registration which is no problem) as follows: "... (c) manual three-dimensional definition of a mask encompassing the area of high signal intensity on DW images followed by automatic three-dimensional stroke segmentation (Functional Magnetic Resonance Imaging of the Brain automated segmentation tool) (briefly, this corresponded to segmentation of the DW voxels within the mask into six gray-level classes, keeping only voxels of the two highest classes)..."
>
>
> [cid:b0542797-8ac5-47e2-bb0d-a6408575d5d5]
>
>
> This seems like a pretty good idea - by defining a mask roughly encompassing the stroke area and constraining the FAST segmentation to this region one could exlude all other "bright voxels" in the brain from inclusion into the segmentation.
>
>
> Maybe anyone has an idea on how to accomplish this?
>
>
> Thanks a lot and best regards,
>
>
> Matias
|