Enabling the extraction of fMRI motion correction parameters


#1

Hi,

We have started a pull request to implement useful outputs from the sct_fmri_moco function, such as motion parameters to input into physiological noise modelling frameworks.

Any suggestion is welcome!

Julien


#2

Hi @jcohenadad - Sorry for the delayed reply… just saw this post. I have observed that fMRI motion correction parameters can be unreliable, and I always use them with caution. Thus, a helpful option to protect against sporadic erroneous ‘correction’ parameters would be a 3- or 5-point moving median filter that may be applied to one or more of the parameter outputs from motion correction algorithms. I found this to be very helpful with my (mainly 7T) data to protect against sporadic field fluctuations masquerading as rigid-body motion.

Cheers,

-Rob


#3

Thank you @barryrl,
good idea indeed. We had it in the past, and then removed it because it was giving spurious regularization on some time series with funky motion, but this is definitely something we could make optional again. Since this is not directly related to the motion param output, i’ve opened a specific issue.
Cheers,
Julien