-----Original Message-----
From: SPM (Statistical Parametric Mapping)
To: [log in to unmask]
Sent: 6/22/2006 10:44 AM
Subject: Re: [SPM] RPV image - is it used?
As an additional note:
The RPV is generally oddly scaled (it appears dark with an
auto-window) and can be hard to examine. I always transform the RPV
image to a FWHM image with ImCalc; the relationship is
FWHM = RPV.^(-1/3)
This creates an image showing a local FWHM estimate. The FWHM image
is useful for gauging the degree of nonstationarity and can even hint
at outliers and artifacts.
-Tom
-- Thomas Nichols -------------------- Department of
Biostatistics
http://www.sph.umich.edu/~nichols University of Michigan
[log in to unmask] 1420 Washington Heights
-------------------------------------- Ann Arbor, MI 48109-2029
On Thu, 22 Jun 2006, Satoru Hayasaka wrote:
> Matthew (and SPMers),
>
>> A friend asked me about the RPV image and its use in statistical
>> thresholding. I did a very quick scan of the SPM2 and SPM5 code, and
>> couldn't see where it was being used. Did I miss something obvious?
>
> The RPV (RESEL-per-voxel) image is a byproduct of spm_est_smoothness.m
> function. This function calculates the average smoothness, to be used
as
> one of the parameters in the multiple comparison correction in SPM. As
a
> part of the intermediate steps, this function spits out the RPV image.
> Although the RPV image per se is not used in the thresholding process,
> the information in the RPV image is averaged and used to determine the
> threshold.
>
> Hope this helps.
> -Satoru
>
|