3dDeconvolve

AFNI Message Board
1

Hello AFNI group,

I have already posted the message below but did not get an answer yet, so I try to re-post … sorry about that.

I am working with brain vascular reactivity data (CVR) and analyze our data using AFNI.
We use changes in CO2 as our stimulus file and acquire data using BOLD imaging.
We usually do a simple linear regression of BOLD by the end tidal CO2. (3dDeconvolve -mask CVRmask+orig -input CVR+orig -polort 1 -num_stimts 1
-stim_file 1 pco2_0.1D -stim_label 1 pco2effect -bucket CVR_slope )
We also have a script to convolve the end tidal CO2 with an exponential of variable time characteristic to account for variable speed of vascular reactivity depending on location and diseases.
This works pretty well to account smoother BOLD responses in impaired vasculature …

However, I would like to be able to recover the HRF using 3dDeconvolve with the BLOCK or TENT function , with something like:

3dDeconvolve -mask CVRmask+orig -input CVR_raw_scaled+orig -polort 1 -num_stimts 1
-stim_file 1 CO2.1D -stim_label 1 CO2effect
-stim_maxlag 12
-stim_times 1 tname BLOCK(1,1) …

The problem is that I cannot make a tname file because the CO2 is not only ON(1) or OFF(0) but rather a graded stimulus with values between 0 and 1 …

Any suggestions/ help?

Thank you,

Julien

2

Do you want to find a different HRF at every input voxel (or ROI)? Sort of what TENT does, but for a continuous time series as the stimulus input, rather than a discrete stimulus time series?

3

If the above is the case, then I figured out how to do this with 3dDeconvolve, combining the -stim_times_AM1 option and the TENT(b,c,n) response model. Although I wrote this code, I never thought of using it this way.

4

Thank you very much for your reply.
Yes, it is exactly what I want to do :slight_smile:

5

Hi Robert W. Cox,

"If the above is the case, then I figured out how to do this with 3dDeconvolve, combining the -stim_times_AM1 option and the TENT(b,c,n) response model. Although I wrote this code, I never thought of using it this way. "

Thank you, so I should be able to do it with those 2 options, right? I looked at -stim_times_AM1 and see how it could works … but will have
to understand it better and try it.
I’ll let you know, thanks!

Julien.

6

I will write up a detailed description of an example, and post a link to it – tomorrow (Thursday 11 Mar 2021)

7

Try reading this
https://docs.google.com/document/d/1ksnsT8y0a-nDasdHlfZi4ytVKrfoJ-TZff-rZXG6nuM/edit?usp=sharing
and see if it is helpful.

8

Thank you so much!!
I did not open the link yet but I’ll let you know after I try.

9

I have just read your doc. It helps a lot and clarifies the use of -stim_times_AM1.
Thank you very much to write that in such a short time!
I will try it on our data in a week or 2 …
Julien.