Modifying BOLD sampling length

[leotozzi88]

Dear experts,

Thank you very much for this great software!

My situation is as follows: I have data collected through multiband fMRI and I would like to build a model where I simulate the BOLD signal and then compare it to the true signal. Ideally later I would like to optimize the model hyperparameters as the ones giving me the least error between the simulated timeseries and the real one in a region of interest.

My problem is that my BOLD signal is sampled every 0.71 s (1.41 Hz), whereas the bold generated from the models is downsampled to 0.5 Hz. I would like to get the BOLD from the model downsampled at the same sampling rate as my true signal, so I can compare them.

One solution of course could be to downsample my signal to 0.5 Hz, but I was wondering if there is a model parameter where I can explicitly set the BOLD downsampling to match my true sampling rate.

Thank you very much!

[caglorithm]

Hi Leonardo! Changing the sampling rate of the BOLD signal is possible with neurolib, however, we haven't yet implemented a dedicated parameter for it.

You can go here in Line 17 of neurolib/models/bold/model.py and change the sapling time from 2000 to 1408 which corresponds to 1000/1408s = 0.71 Hz and should give you the desired sampling rate.

Hope I could help!

[leotozzi88]

Thank you very much!