Artefact correction

[thomas.nickson]

Does Nipype (or nipy) offer any way of correcting artefacts such as interpolation or replacement by the mean image?

I've used rapid art to detect problematic areas but not sure how to go about correcting them. Does anybody have such a script in the works or a place where I could read about methods and perhaps implement one myself?

Thanks.

[Satrajit Ghosh]

hi thomas,

we simply regress the volumes out (one column per outlier with a 1 at the volume that's bad) these regressors don't get convolved with a hrf. this effectively reduces that volume to a mean (although i don't think i have ever proved that). the specifymodel interfaces allow an input for outliers.

interpolation becomes hard as that would depend on the frequency of outliers - but there are techniques to potentially impute missing values - would be a research problem - and depend on the assumptions made about the characteristics of the signal.

cheers,

satra

--
 
---
You received this message because you are subscribed to the Google Groups "NiPy Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nipy-user+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Thomas Nickson]

Thanks for the helpful reply as usual Satra. I wonder if it's possible to do this without using glm for experiment modelling such as in resting state. Simply using the regression to get a volume mean?

Thanks,

Tom

[Satrajit Ghosh]

hi tom,

this is essentially a nuisance regression glm, where all covariates are nuisance. simply use something like FSL_GLM or other such tool as a glm - you don't need any hrf convolution here. furthermore you are interested in the residuals from such a glm as your cleaned data.

cheers,

satra