Question Posted 06/21/16:
Dir Sir or Madame,

Currently, I am working with PET scans from the ADNI database in my research.

I try to understand all of the four pre-processing types ( I would like to know more about the intensity normalization performed in the third pre-processing type. Unfortunately, I cannot find further information about how this is done in detail, except that the average of all voxels in a mask should be equal to one.

I would like to ask if you could tell me how this subject-specific mask is generated in particular.

Yours sincerely,

Markus Hubrich
Response posted 06/21/16 by Robert Koeppe:
The answer is different for FDG than for the other tracers, PiB, AV45, AV1451.
For FDG, the global mean is normalized to a value of 1.0. This intensity scaling isn't meant to be used as the final scaling values. One is free to re-normalize to any reference one wants. The other tracers are all normalized to the cerebellar gray matter, but again, if one wanted to normalized to a different region, (pons, white matter), these processed sets can be renormalized to whatever region you want. If you have further questions, please e-mail me at
Response posted 07/04/16 by Markus Hubrich:

Thanks for your answer!

Currently, I’m working only with the FDG subset.

Could you specify the term “global” mean normalization? In my mind, there are two different interpretations, depending on how one understand the term “global” here:
1. Each scan is normalized to have mean equal to 1.0.
2. All scans are scaled so that the mean over the entire FDG subset is equal to 1.0.

I would be glad to hear which of those two options is true.

My best regards,