Hi afni experts，

I am interested in the superior colliculus (sc) and sc has an optimised HRF with a 4s peak (Wall, M. B., Walker, R., & Smith, A. T. (2009). Functional imaging of the human superior colliculus: an optimised approach. Neuroimage, 47(4), 1620-1627.). The optimised HRF used in the reference is the standard HRF which is a synthetic function produced by combining two gamma functions (Friston et al., 1998; Handwerker

et al., 2004) and is implemented in BrainVoyager QX.

I don’t know how to use the same combined two gamma function with a 4s peak as HRF in afni. Though I notice the ‘GAM(p,q)’ option of Rmodel in 3dDeconvolve and the latency can be changed by changing the value of p*q. But the expression used in this option is just a single gamma function.

So :

- How to use the same synthetic function with a 4s peak as HRF in afni?
- If it is too complicated to use the synthetic function mentioned in the reference, ‘GAM(p,q)’ can also be accepted. But how to set the two parameters ‘p’ and ‘q’ properly? Could ‘p’ and ‘q’ be set arbitrarily to satisfy the equation p*q=4 if I need a 4s peak HRF? For example, p=2,q=2; p=8.6,q=0.4651 or p=7.3126,q=0.547. Are these combinations of ‘p’ and ‘q’ all OK? What do the two parameters (p & q) model in the gamma function? The magnitudes of the peak, time-to-onset or something else?

Thank you in advance!