Re: [Freesurfer] Mean intensity with mri_segstats : different results with .annot or .label

You need to weight by the number of vertices

n = [27805 2321 552]; m = [8.8194 10.3661 10.3365]; sum(n.*m)/sum(n)

ans =

8.9637

On 11/10/2016 06:44 AM, Matthieu Vanhoutte wrote:

Dear Freesurfer's experts,

Could anyone please explain me the difference I got with command line in below mail ?

Best regards, Matthieu

2016-08-12 12:07 GMT+02:00 Matthieu Vanhoutte <matthieuvanhoutte@gmail.com mailto:matthieuvanhoutte@gmail.com>:

Dear experts,

I am in trouble with two ways of computing mean intensity with
mri_segstats.

First I have used on .annot files with three different labels
inside (SegId 1 to 3) :

/mri_segstats --annot fsaverage lh cache.th23.pos.sig.ocn.annot
--i lh.PET.fsaverage.sm10.mgh --sum lh.pet.sum/
/
/
which give me those results:
# ColHeaders  Index SegId NVertices Area_mm2 StructName Mean
StdDev Min Max Range
1   1     27805    14182.6  cluster-001      8.8194     1.4351
6.2105    14.2947     8.0842
2   2      2321     1047.3  cluster-002     10.3661     1.0848
8.8375    12.9317     4.0942
3   3       552      313.1  cluster3      10.3365     0.4711
9.3719    11.8694     2.4975

Second I have created on .label files containing the three
segmented labels (SegId 1 to 3) and used this label with mri_segstats:

/mri_segstats --slabel fsaverage lh
lh.th23.cluster.thresh1.3.label --i lh.PET.fsaverage.sm10.mgh
--excludeid 0 --sum lh.pet.label.sum /
/
/
which give me those results:/
/
/# ColHeaders  Index SegId NVertices Area_mm2 StructName Mean
StdDev Min Max Range /
/  1   1     30678  15542.9  Seg0001 8.9637     1.4701 6.2105  
 14.2947 8.0842 /
/
/
I verified if I obtained the same results with the two ways
(.annot or .label) and everything is fine comparing NVertices,
Area_mm2 but Mean in first way gives me (8.8194+10.3661+10.3365)/3
= 9.84066 and with .label file Mean = 8.9637.

Did I have miss anything and how is computed mean intensity ?

Many thanks in advance.

Best regards,
Matthieu

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer