Hi Renata,
In general, mixing’n’matching is not a good idea, since different sequences bias the volumes in different directions. Now, if the differences in sequences were independent of whether subjects are patients or controls, it wouldn’t be *that* bad. But, if patients have different sequences than controls, the results are catastrophic because you cannot tell whether detected differences are due to anatomy or to the differences in sequences … 
- For the T1s, rather than having two groups, you could possibly resample the higher-res (0.8) images to 1mm isotropic. But I wonder if this is introducing a bias (I’d expect it to be small, but…). 
- T2 vs FLAIR: I’d need to see examples to decide. In principle, the one with higher contrast and resolution is better ;-)
Cheers,
/Eugenio


Juan Eugenio Iglesias
Translational Imaging Group
University College London
http://www.jeiglesias.com
http://cmictig.cs.ucl.ac.uk/


On 30 Aug 2016, at 17:28, Vaz pandolfo, Renata <rvp8@njit.edu> wrote:

Dear Eugenio,

I have a couple of questions regarding the hippocampus subfields segmentation on the dev FS 6.0 which I hope you could help me with.

We are comparing the volumes of hippocampal subregions of patients and controls, and we would like to be as consistent as possible with the input so that the output reflects only the differences we expect to see.

The problem is that we have a large number of controls but not all of them have the same sequences (T1, T2 or FLAIR) as the patients. So our questions are:

If the T1s from some subjects have a 1x1x1 resolution and others have 0.8x0.8x0.8 isotropic resolution, do we have to create two groups of controls and patients for each sequence? In other words, does the resolution of the T1 affect the freesurfer segmentation in a way that we can't compare 1x1x1 with 0.8 x 0.8 x 0.8?

Also, we would like to use T2s or FLAIRs as additional inputs to improve the results. Which sequence is preferably recommended? In our case, most controls and patients have the same FLAIR sequence, so to keep it consistent we would probably use FLAIR if that's worth it.

Finally, we also have different T2s: some isotropic, some anisotropic and partial cooverage of the brain, and some anisotropic that are aligned to the hippocampus (resolution: 1x1x3, best resolution on coronal slices). Considering that they are just additional inputs, do they have to be consistent as well? In other words, should we only compare patients and controls that had the same T2 sequence as additional input and same T1 sequence as main input, or can we "mix and match" the additional inputs?

Thank you in advance for your help,

Renata
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.