Closed behnam12 closed 6 years ago
Hi @behnam12 ,
it is very much expected that if you use different models, then the estimated parameters are different. In your case, which local model did you use with COMMIT? In fact, COMMIT is not a model, but a framework to combine a local model with fiber tracts. Also, your NODDI map looks a bit weird, like the values are saturated.
Best, Ale
Hi Alessandro ,
Yes, you are absolutely right about difference between estimated parameters of COMMIT and NODDI. I just think why there is a border between middle part and other parts in my IC map. I used Stick-Zeppelin-Ball model.
Best Behnam
I don't know well the anatomy there, but if there are two bundles that kiss or cross at small angles (so, they can be compact), then the IC map can be reasonably higher than the rest. I am more concerned about the saturated values of the NODDI maps, this does not look realistic at all!
Hi Alessandro ,
Thank you for your explanation about IC map from COMMIT and NODDI results. You are right about my NODDI results. I just want to add supplementary information about my IC map from COMMIT. As it can be seen from the figure, IC values vary for different diffusivity parameters. It could be due to fixation artifact. In fact, non-uniform fixation across tissue leads to spatially varying diffusivity.
Thanks for providing this great framework.
Best Behnam
Hi @behnam12 ,
if you change the intrinsic diffusivities of the model you use to fit, then the estimated maps (e.g. ICVDF) will be different. In theory, this should be estimated in each voxel. But, as done by most models to reduce complexity and increase the robustness of the fit, you fix it and you have to set to the most appropriate value for your experiment. In your case, it's ex-vivo tissue, so the first plot is completely off (saturated). Not sure, though, of the spatial dependency of the fixation as I'm not an expert there. This could be an explanation, but it could also arise from an actual/real configuration of the axons in that middle part of the sample.
Hi Alessandro ,
Thank you for your comprehensive explanation.
Best Behnam
Hi COMMIT group,
I used COMMIT for ex-vivo optic chiasm dataset. I just wondering why middle part, region containing crossing fibers, has higher value that other parts ? In fact, I expect to get the same result that I got from NODDI for intracellular compartment map.
Thank you Behnam