According to the paper: "The amount of regularization (lambda) varied by a
factor of ten so that ten times as much regularization was used for high
regularization as was used for medium regularization, and ten times as much
regularization was used for medium regularization as it was for low
regularization." So I guess if Medium = 1, Low = .1 and High = 10.
Hope this helps,
Cameron
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of John Ashburner
Sent: Thursday, August 07, 2008 10:43 AM
To: [log in to unmask]
Subject: Re: [SPM] Normalisation using Unified Segmentation in SPM5
> Sorry, my mistake, I was aware about this difference. However, I'm still
> not sure whether this is the correct procedure: 1. "Coregister" T1 of the
> subject to the EPIs of the subject
> 2. "Segment" the T1 of the subject including normalisation
> 3. "Normalise" the EPIs by applying the matrix generated by step 2
That's it. When you do the coregistration, there is no need to do any
reslicing.
>
> > This paper may be of interest here:
> > Crinion J, Ashburner J, Leff A, Brett M, Price C, Friston K. (2007).
> > "Spatial normalization of lesioned brains: Performance evaluation and
> > impact on fMRI analyses". NeuroImage 37(3):866--875
>
> Thanks a lot, this paper was very helpful and informative
>
> A question regarding this paper: The regularisation manipulated in the
> paper was the "Bias Regularisation" in he SPM5-Segment options, is that
> correct? If that is right, which are the according regularisation levels
in
> SPM5? Light-Medium-Heavy?
It was the warping regularisation, but I don't remember what values Jenny
used. I think "medium" was the default value of 1, but I'm really not sure
of the others.
Best regards,
-John
|