Simone,
Opinions may vary, but I do not think that the smoothness
estimate that SPM produces is equvalent to the true scanner
resolution.
The best way IMHO to measure your PET scanner resolution is
to run a point source phantom in a water matrix and measure
the FWHM based on your given reconstruction paramers. In
Pittsburgh we found the true reconstructed PET camera
resolution of an ECAT HR+ in 3D to be 7mm x 7mm x 4mm based
on a Hanning cutoff of 0.4 I think. I would have to look it
up I am going off of memory right now. see:
Meltzer CC. Kinahan PE. Greer PJ. Nichols TE. Comtat C.
Cantwell MN. Lin MP. Price JC. Comparative evaluation of
MR-based partial-volume correction schemes for PET. [
Journal Article] Journal of Nuclear Medicine. 40(12):2053-65,
1999 Dec.
I think the point source phantom is in there.
hope this helps.
Phil Greer
UPMC MR and PET Center
Pittsburgh PA
-----Original Message-----
From: [log in to unmask] <[log in to unmask]>
To: [log in to unmask] <[log in to unmask]>
Date: Thursday, July 12, 2001 9:53 AM
Subject: FWHM
>Dear SPM-ers,
>
>For our single subject PET data we are interested in the resolution of our
>'raw' images (e.g. camera resolution). We tried to obtain this raw FWHM by
>doing a statistical analyses on the images which are realigned but not
>normalised or smoothed (avoiding the introduction of extra smoothing by one
>of them) and read the FWHM from the spm output.
>
>As always, during the realignment, we created a 'mean image only'(I)
>resulting (during the statistics) in the following error:
>
>---
>SPM99: spm_spm_ui (v2.26) 15:40:05 - 09/07/2001
>========================================================================
>Mapping files : ...done
>??? Error while evaluating uicontrol Callback.
>
>
>>> ??? Error using ==> spm_spm_ui
>images do not all have same orientation & voxel size
>---
>
>We speculated that the statistics needs the information from the image
files
>and does not bother about the *.mat files (containing the orientation and
>voxel size information). So we created during the realignment the 'mean
>image + all images'(II) and used the r*.img files for the statistical
>analyses. No error messages occurred and we found an FWHM of 5.0 5.7 5.6 .
>
>However, for the normal statistical inference we identically normalised and
>smoothed (10mm) the two different sets of realigned images (I and II).
>
>Now we find in situation (I) a FWHM of: 9.5 10.4 10.8
> and in situation (II) a FWHM of: 9.6 10.6 11.0
>
>Of course this is a very small difference. But the question is: is the FWHM
>of 5.0 5.7 5.6 our raw camera resolution. Or is here already some
smoothness
>introduced by the realignment (which is suggested by the results after the
>normal statistical analyses)? If yes, does anyone know how much?
>
>If the realignment introduces some smoothness, can the raw resolution then
>be calculated by the formula given by Karl Friston:
>FWHM^2 = FWHM{applied)^2 + FWHM{intrinsic)^2
>(http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind9909&L=spm&P=R10782)
>if we knew how much smoothness was introduced in the realignment?
>
>Or is there another way to obtain the raw resolution of the images?
>
>Thanks in advance for your time and help,
>
>Simone.
>
>--
> A.A.T.S. Reinders
> University Hospital of Groningen
> The Netherlands
>
|