Question
Question Posted 03/17/16:
I have searched into documentation but didn't found information concerning resolution FWHM of each PET scanner used in ADNI project.
Do you think it is absolutely needed to use uniform isotropic resolution of 8 mm FWHM PET data for statistical analysis ? Because, in general the more established (though probably less valid way) in the literature is simply to apply the same smoothing kernel to all data irrespective of the scanner (therewith considering the scanner specific smoothness as noise). What is your advice ?
Many thanks in advance !
Could you provide me this ?
Many thanks in advance !
I have searched into documentation but didn't found information concerning resolution FWHM of each PET scanner used in ADNI project.
Do you think it is absolutely needed to use uniform isotropic resolution of 8 mm FWHM PET data for statistical analysis ? Because, in general the more established (though probably less valid way) in the literature is simply to apply the same smoothing kernel to all data irrespective of the scanner (therewith considering the scanner specific smoothness as noise). What is your advice ?
Many thanks in advance !
Could you provide me this ?
Many thanks in advance !
Response posted 03/17/16 by Bob Koeppe:
No, I don't think it is absolutely necessary to use the images with an isotropic resolution of 8mm. However, to use the same smoothing, irrespective of scanner I think is a bad idea. That will cause differential partial volume effects, etc.
So bottom line, I don't think there is any definite right or wrong way to do this. In many cases, the base approach may well be dependent upon the question you are trying to answer (e.g. cross-sectional, longitudinal, across different tracers, etc.)
So bottom line, I don't think there is any definite right or wrong way to do this. In many cases, the base approach may well be dependent upon the question you are trying to answer (e.g. cross-sectional, longitudinal, across different tracers, etc.)
Response posted 03/17/16 by Matthieu:
I would like to compare patients group (at my site) with controls group from ADNI protocol (came from multiple scanners).
I know the resolution of our PET scanner however I don't know resolution of the different scanners used in ADNI protocol for control subjects. Willing to apply partial volume correction, I would need to have resolution of each scanner.
In order to avoid any false statistical results due to differences between scanners, I would be tented to put all patients+control to 8 mm FWHM. But I would lose better resolution of many of my PET images...
I know the resolution of our PET scanner however I don't know resolution of the different scanners used in ADNI protocol for control subjects. Willing to apply partial volume correction, I would need to have resolution of each scanner.
In order to avoid any false statistical results due to differences between scanners, I would be tented to put all patients+control to 8 mm FWHM. But I would lose better resolution of many of my PET images...
Response posted 03/17/16 by Bob Koeppe:
It is hard to say what the resolution of various scanners are for partial volume correction purposes.
I can provide a list of how much smoothing I applied to reach 8 mm. I agree smoothing to a common resolution of 8 mm is not optimal, because as you said, you are throwing away the better resolution of the newer scanners. Please let me know what type of scanner you have at your site.
I can provide a list of how much smoothing I applied to reach 8 mm. I agree smoothing to a common resolution of 8 mm is not optimal, because as you said, you are throwing away the better resolution of the newer scanners. Please let me know what type of scanner you have at your site.
