Questions about using rfMRI_REST1_AP_hp0_clean.nii.gz for (f)ALFF analysis
jean-simon fortin
Dear HCP team,
I am planning to run f/ALFF analysis on a subcortical structure (the habenula). For this, I am working with the following image:
I have a few questions to clarify before proceeding:
-
My understanding is that for fALFF we should use unfiltered data. Has this hp0_clean image been high-pass filtered, or does “hp0” indeed mean that no high-pass filter has been applied?
-
My understanding is that the same unfiltered data can also be used to compute ALFF, since ALFF is calculated within the 0.01–0.08 Hz band and would therefore not be affected by high-pass filtering. Is that correct?
Has the rfMRI_REST1_AP_hp0_clean.nii.gz image already been detrended? Based on Woletz et al. (2018), Beware detrending: Optimal preprocessing pipeline for low‐frequency fluctuation analysis, polynomial detrending can be beneficial for ALFF. I was planning to apply polynomial detrending to the hp0_clean image. Would that make sense, or has detrending already been performed by HCP?
For fALFF, the same paper suggests that polynomial detrending should be avoided. However, the toolbox I am using requires me to select a polynomial trend order between 0 and 3. Would it be appropriate to choose polynomial trend=0, given the preprocessing already performed by HCP? My understadning is that by choosing polynomial trend=0, I would avoid the negative consequences of polunomial detrending identified in the paper. However, I'm wondering if choosing polynomial trend=0 would be redondant with HCP preprocessing and if it could cause issues. Thank you very much for your guidance. Best regards, Jean-Simon
Glasser, Matthew
I would use the new data we have just released on ConnectomeDB powered by BALSA. If you need the habenula, that is not currently in the CIFTI grayordinates space, so you would have to use the volume timeseries. We would recommend: ${StudyFolder}/${Subject}/MNINonlinear/Results/rfMRI_REST/rfMRI_REST_hp0_clean_rclean_tclean.nii.gz.
With regard to the processing, hp=0 is just a linear detrend, which is fine. We don’t do temporal filtering. If you are considering polynomial detrending, you would need to have some phantom data showing that there is actually a significant nonlinear component of gradient heating. We have seen this for very long continuous scans (like 1 hr), but shorter runs are going to be piece-wise nonlinear, and it won’t help, but certainly could hurt.
I am not super familiar with the details of ALFF/fALFF, but we have an alternative approach that focuses on the amplitude of the BOLD neural signal (but does not make frequency assumptions about the BOLD signal) that we will be making available in the future.
Matt.
--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
hcp-users+...@humanconnectome.org.
To view this discussion visit
https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/d4a86508-564d-47ce-aba1-dfdbb3f10c03n%40humanconnectome.org.
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.