3DDeconvolve: Insert a regressor to be convolved with response function

I would like to enter a regressor that represents the neuronal activity (acquired with electrophysiology), therefore. a hemodynamic response function should be fitted to it.
It is a time series with the same duration as the fMRI scan consisting of continuous values.

I can’t find a way to do it in 3dDeconvolve: -stim_times is expecting onset times in seconds. -stim_file would accept a time series, but does not get convolved with the response model, and neither does -ortvec.

So is it the best option to manually convolve my regressor in matlab and then enter it as -stim_file? That’s ok, but I would have liked to use one of the variable shape models in 3dDeconvolve.

Not quite sure exactly what you’re looking for. Suppose that the neuronal activity at 3 consecutive time points are a1, a2, and a3. Do you want to obtain a regressor as the summation of three consecutive HRFs with each modulated by a1, a2, and a3?

Yes, basically I want the HRF, which was fitted to the data not from a stimulus time series (zeros and ones) but from the neuronal signal that has some continuous value at every time point.

-stim_times_AM2 might be exactly what I want, if I define every single time point as stimulus event which is amplitude modulated by the neuronal signal. Will try…

Depending what you’re looking for, -stim_times_AM1 might be another possibility.

Use “waver -FILE” for that. The convolution file
that you create can have the desired waveform,
such as GAM or BLOCK(1).

This is done in the sample PPI analysis under
AFNI_data6/FT_analysis/PPI/cmd.ppi.2.make.regs.

Search for “waver”.

  • rick