Quantcast
Channel: NITRC CONN : functional connectivity toolbox Forum: help
Viewing all articles
Browse latest Browse all 6861

False B-estimates confound-corrected timeseries

$
0
0
Dear Alfonso,

I am running into a problem when using the confound-corrected timeseries for first-level analysis in SPM.
I have preprocessed and denoised my data using:

- bandpass filter 0 - Inf
- no detrending
- despiking after regression (in retrospect, I think I could have set this to no despiking because of the use of scrubbing)

My list of confounds includes WM, CSF, scrubbing and realignment parameters (I have removed the session-specific conditions from the list of confounders).
First level analysis on the confound-corrected timeseries yields extremely high beta estimates of up to 900, as opposed to ~10 when using the dataset prior to denoising.I have set the global normalisation to 'none' in SPM. I also tried setting the BOLD signal units to raw in the CONN options list, but this gives a similar result.

Do you know what could cause this kind of problem?

Best regards,
Bram

Viewing all articles
Browse latest Browse all 6861

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>