Hi AFNI experts,
We have a multi-echo dataset, and we want to use it for MVPA analysis. Therefore, we need to keep each subject's data unnormalized and unsmoothed. However, it seems that meica.py performed all these preprocessing before ICA denoising (https://afni.nimh.nih.gov/pub/dist/src/pkundu/README.meica). I am wondering if it is reasonable and possible to skip the normalization and smoothing steps and run the ICA denosing step directly?
Thanks in advance,
I believe the answer is yes: it is possible to perform the echo dependence analysis prior to any smoothing and spatial normalization steps.
ME-ICA has two main programs meica.py and tedana.py. Meica.py is designed to perform some preprocessing in the data and then conduct the echo-dependence analysis using tedana.py.
Users could skip using meica.py and directly run tedana.py in their data at whichever point in the pre-processing they consider.
Now the question of when (in the pre-processing pipeline) it is optimal to run the echo-dependence analysis step remains open I believe.
Many thanks for your helpful reply! I will try tedana.py.
All my best,
Also, the current defaults for meica.py are for no smoothing, and it will only warp the dataset to standard space if the option --MNI or --space are used. It’s likely that just using meica.py as is will work for your purposes (the input to tedana.py must be a spatially concatenated dataset which meica.py will generate for you). If not, you can use the --script_only option to output a script that you can then modify to your needs.