BalloonModel defaults

[Lia Domide]

Hi,

I got a questions/possible bug report from a TVB user:

I looked at the implementation of the BalloonModel in scientific_library/tvb/analyzers/fmri_balloon.py and I think there might be an error in the default values tau_s and tau_f. I'd like to open an issue for this.

In equation (7) from [1] the differential equation for the signal component is:

ds/dt = epsilon * u(t) - s / tau_s -(f - 1) / tau_f

The default values chosen here are: epsilon = 0.5, tau_s = 0.8, tau_f = 0.4

Paper [2] builds upon paper [1] and uses the same set of differential equations, equations (3). Here the signal component is given as:

ds/dt = z - kappa * s - gamma * ( f - 1)

Here u has been renamed to z, epsilon is dropped, and instead of tau_s and tau_f they now use kappa and gamma. They act as the inverse of these time constants. According to table 1 the default values are kappa = 0.65 per s, gamma = 0.41 per s.

So in paper [2] the values are:

tau_s = 1 / kappa = 1 / 0.65 s ~ 1.54 s
tau_f = 1 / gamma = 1 / 0.41 s ~ 1.44 s

The issue I am raising is that the default values of tau_s [3] and tau_f [4] in the tvb implementation are:

tau_s_tvb = 0.65
tau_f_tvb = 0.41

To me it seems as if tau_s and kappa and tau_f and gamma got confused with each other and the default values should be the inverses of what they are now.

[1] Friston et al., 2000, Nonlinear Responses in fMRI: The Balloon Model,
Volterra Kernels, and Other Hemodynamics

[2] Friston et al, 2003, Dynamic causal modelling

[3] https://github.com/the-virtual-brain/tvb-root/blob/d50008bd7c81e1844f7ecbba1fc59fc542e0a326/scientific_library/tvb/analyzers/fmri_balloon.py#L127

[4] https://github.com/the-virtual-brain/tvb-root/blob/d50008bd7c81e1844f7ecbba1fc59fc542e0a326/scientific_library/tvb/analyzers/fmri_balloon.py#L133

I think he is using the web GUI, where these parameters are not exposed. So I would like to propagate this question towards this group with a double goal:

- is this indeed a mistake in tvb code?

- should we expose all params in web GUI (would it make sense) ?

Thanks,

Lia.

[WOODMAN Michael]

Hi

This does look like a mistake in the default parameters used for the BOLD analyzer, not a mistake in the code per se.  The effect would be to make the BOLD response a bit faster than it would be if the parameters given by Friston’s papers were used directly.

Exposing each parameter in the UI would be the most generic option, and allow the user to tweak according to the paper they prefer citing for BOLD modeling.

Cheers,

Marmaduke

On 31 Aug 2020, at 14:04, Lia Domide <ldo...@gmail.com> wrote:

Hi,

--
You received this message because you are subscribed to the Google Groups "TVB Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tvb-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/03e8dc56-5aeb-4695-9e82-8eaed2c22731n%40googlegroups.com.