Using TEDPCA and preprocessed fmriprep data

Hello All
New to using Tedana
I was wondering if it was possible to apply on the PCA component of the tedana application on a set of preprocessed fmriprep data

in Python I have imported:
import tedana
from tedana.decomposition import tedpca

Would anyone have suggestions or examples of input parameters
Thanks so very much.

Hello @Azeezat_Azeez,

Could you go into a bit more detail about what you want to do with the PCA? In tedana, the PCA step is just used to reduce the dimensionality of the data to produce a better ICA fit. The following ICA is the primary denoising step in tedana, so most folks wouldn’t want to run the PCA on its own.

Hello @tsalo
Thank you for relaying
I have data that has already been ‘cleaned’
I just wanted to use the PCA on the cleaned data
I see on the documentation there is a API called tedana.decomposition.tedpca
Wondering how I could use this on ‘cleaned’ data(inputs/outputs)
Hope this makes sense
Thank You,

I don’t completely follow your question because I’m not sure what cleaning steps you’ve done. If you’re using fMRIPrep you can insert tedana into the processing pipeline.
There’s a bit more nuance, but the general rule is you want to apply the same motion correction and registration transforms to all echoes and you don’t want to do any preprocessing steps that will affect the relationships between the echoes. (i.e. Don’t mean scale the data or do any agressive smoothing or other norming steps).

I hope this helps

Apologies for the confusion
I would like to run tedana after fmripreped (21.0.1)data that has already combined echos
I have tried python from the docs page it does not seem to
Any advice would be greatly appreciated
Thank You

tedana requires the separate echoes as an input because the method compares data across echoes as part of the denoising process. I’m not an fmriprep user, but I assume the preprocessed separate echoes from right before the combination step are saved by fmriprep. @tsalo is more of an fmriprep user and might be able to tell you exactly which file names to use. You should be able to take those file names as use them as an input to tedana.

Hi Azeezat. If you’re using 21.0, you should be able to add the --me-output-echos flag to get outputs that are usable with Tedana.