Thank you Cyril and Hedok for really helpful answers!
However, I still feel some uncomfortable to my data. I have two group subjects (patients and controls). I want to find the difference between them. However, for there are a lot of 1 in the grey matter data, I find the most salient differences are usually located in the edge of gray matter. Is it correct?
Need I do some blur any more?
If it is needed, how to do it?
Thanks, again!
Xiaochu
-----Original Message-----
From: Hedok Lee [mailto:[log in to unmask]]
Sent: 2008年1月28日 10:56
To: [log in to unmask]
Subject: Re: [SPM] The voxel value in VBM data
Dear Xiaochu:
Yes, numbers should be between 0‾1, but it may exceed one for modulated
images("mw" images)
From what I recall, a standard VBM analysis outputs native("c" image),
unmodulated("w" image), or modulated("mw" image) for each tissue
compartment. In native and unmodulated images, segmented image
intensity represents probability of belonging to a particular tissue
compartment, hence the sum of the three compartments, GM+WM+CSF, should
always be one.
In the case of modulated image, there is an additional scaling called
Jacobian determinant to account for how much brain volume has
expanded(greater than 1) or shrunk(less than 1) to match a template. In
this case, image intensity may exceed one because (GM
probability)*(Jacobian determinant)>1. This is how I understand the
difference between modulated and unmodulated schemes.
Hope this helps,
Hedok
Zhang, Xiaochu (NIH/NIDA) [F] wrote:
>
> Hi Spmers,
>
> I am working on a VBM study. When I got the result, I found the value
> of each voxel was confused me.
>
> The value of all voxels are from 0 to 1. Is it correct?
>
> Additionally, a lot of voxels within grey matter were 1. Is there a
> ceiling effect here? How to change the scale of it?
>
> Cheers,
>
> Xiaochu Zhang Ph.D
>
> Visiting Research Fellow
>
> NIH/NIDA-IRP
>
> 5500 Nathan Shock Drive
>
> Baltimore MD 21224
>
>
>
> Tel: (410) - 550 - 1440 ext. 434
>
>
>
