Hi everybody (and I guess Marko Wilke and Christian Gaser if you're reading this) !
I have been trying to use the cerebr-o-matic toolbox to obtain tailored TPM for my developmental sample to use to perform segmentation with the CAT12 toolbox.
Some words on my sample: I have 100 children between 8 and 12.5 years. They belong to three groups: typically developing children; dyslexic and ADHD. I want to compare these groups in terms of different brain characteristics, including cortical thickness and GMV in a specific set of regions.
I decided to take some time to find out what it can be the best approach to segment and obtain cortical thickness and GMV for this sample. To this end I wanted to compare the final results (i.e. segmented images) using (1)SPM12 with the standard prior, (2)using prior from the COM toolbox using the provided parameters and (3)using prior created estimating my own sample parameters (including in the model age, sex and diagnosis).
The process of creating the prior with the COM toolboxes worked fine: both using the parameters provided with the toolbox and estimating my own sample parameter led to the creation of a 6D volumes that look fine (except for the fact that the 6th volume - that if I got it correctly model the background - of each 6D volume has zero everywhere).
However, when I try to use these priors with CAT12, I got two different error messages:
-when using the prior calculated using the provided parameters I got an error saying :
"WARNING: Inverse tissue contrast!
(BG=5.40, CSF=1360.58, GM=578.46, WM=159.68)
AMAP estimated untypical tissue peaks that point to an
error in the preprocessing bevor the AMAP segmentation."
-when using the in-sample parameters I got an error saying:
"No largest WM cluster could be found:
Please try to set origin (AC) and run preprocessing again
because it is very likeli that spatial normalization failed."
Note that when I try to process the same T1 images using the standard SPM12 prior everything run pretty smooth.
Do you have any idea about what is going wrong ? Is there something obvious that I am missing ? Is the culprit the fact that the background prior is uniform (=0) the culprit ? And is there a way I can fix this in the COM pipeline ?
Thanks in advance for any help you can provide
Best
Alain
|