Hi - here's in more detail how to get agreement between avwstats and MEDx:
I have an image which gives:
avwstats structural_brain_pve_0 -V -M
353422 706844.000000 0.389268
so this gives me total volume for this class of 275151.75 mm^3
In MEDx I can get the same result easily
a) the easiest is to take find the SUM and * by voxel size
gives 275252
b) the more equivalent way is to set a very low lower threshold (eg 1e-15)
and the multiply MEAN and AREA (ie volume)
gives 275151.6
Thus there is no problem here. What may be the source of your confusion is
what we mean by totalling up partial volumes. For each voxel, a value is
between 0 and 1 for the partial volume fraction of the class of interest.
say this is v_i at voxel i. Thus the total volume for this class
contributed by this voxel is v_i * V (where V is voxel volume in mm^3).
Thus if we want the total volume for this class we want
Sum_over_non_zero_voxels ( v_i * V )
=
V * number_of_non_zero_voxels * mean_of_v_i_over_nonzero_voxels
=
[volume of nonzero voxels] * [ mean of nonzero voxels]
Hope this is clearer now :)
Stephen M. Smith
Head of Image Analysis, FMRIB
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
On Fri, 17 Aug 2001, Qingyan Guan wrote:
> Hi,
>
> Thank you very much for your comments.
>
> I re-calculated the lesion volume using both FSL and MEDx. below, I used
> only one data set as comparing data.
>
> I compared the volume value by using same partial volume maps.
>
> Using FSL avwstats program for partial volume maps(lesion)
> (*_brain_pve_0.img) segmented by FSL.
>
> volume= 152450.286 (mm*mm*mm) *
>
> Using Medx measurement tool for partial volume maps(*_brain_pve_0.img)
> segmented by FSL, adds up the volumes with higher pixel value.
>
> volume(mm*mm*mm) Lthreshold:pixel(mm)
> 160932 0.40
> 155287 0.42
> 152650 0.43 *
> 147495 0.45
> 136155 0.495 **
> 134997 0.5
>
> It shows that value(152650) with pixel value(0.43) is comparable with the
> value(152450) from avwstats.
>
> I wonder if there is a Lthreshold already in avwstats program. Otherwise it
> could be very different.
>
> I also tried using segmentation maps(*_brain_seg.img) that segmented by fsl
> as 4 classes. The 4th class is considered as lesion class. Then add up only
> the 4th class volumes by MEDx.
>
> volume= 136255 (mm*mm*mm) **
>
> This is also a very statiable value, it is not depends on person who did
> the measurement. I think it's supposed to be the same as the avwstats
> value. But why it is so different with it. Could you give me some idea on
> it?
>
>
> additional, I tried using the original brain volumes(E*S*Iall.img), adds up
> the partial volumes with higher pixel value.
>
> volume(mm*mm*mm) Lthresholds:pixel(mm)
> 168745 125
> 152176 127 *
> 137790 129 **
> 131050 130
>
>
> Looking forward to hearing the idea from you. Thanks.
> (If you need the data set that I used, please let me know. I will try to
> send to you).
>
> Best Regards,
>
>
>
>
> QingYan Guan
> [log in to unmask]
> (425)391-8251
>
> _________________________________________________________________
> Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp
>
|