Dwiintensitynorm in aging study

Dear all

I have performed intensity normalization across 2 groups of young (30) and older (27) adults using dwiintensitynorm with default settings. The FA template and the WM mask look fine to me.

However, on average the dwi_norm_scale_factor for young adults is slightly higher than old adults (3 vs 2.55). I was wondering if this is something to worry about. If ‘yes’ how can I fix it?

Thanks,
Hamed

Hi Hamed,

The dwiintensitynorm first generates a white matter mask for the population, and backprojects this to each subject to compute a median white matter b=0 intensity for each subject. The dwi_norm_scale_factor is then applied to all images (b=0 and b>0) of a subject to obtain a median b=0 intensity in white matter of 1000 (default).
The lower scale factor in older adults therefore means their b=0 intensity was higher than in younger adults. The b=0 image is a T2-weighted image and it’s indeed so that older adults can have prolonged (i.e. higher) T2. In that case the algorithm effectively did its job in normalising intensities. The ages (30 and 27) however seem very similar to me, and both are in fact young. Is there any other difference between both that can explain increased T2 values (e.g. lesions)?
It’s important to also rule out differences in acquisition parameters, or a scanner upgrade (in case young and old were not scanned randomly), which can alter the intensities.

Cheers,

Thibo

N.B. if you’re interested in the FA only, normalisation is not required (but not harmful either) as the scale factor is applied to both b=0 and b>0 images, and therefore cancels out in the ratio that is FA.

Hi Thibo

Nice to hear from you and thanks for the explanation.
Sorry I should have mentioned that 30 and 27 are the NUMBER of subjects per group and not AGE.
So I can continue, right?

Cheers,
Hamed

Hi Hamed,

Sorry for my late reply.
If the ages are indeed very different, this may be the reason of different scale factors and it seems normal to me. Just make sure that it isn’t due to some other cause (e.g. acquisition parameters).
Good luck!

Thibo

P.S. I’m also just a user, so all advice is only based on own experiences :slight_smile:

1 Like