Questions about PhysIO -> log_scan_timing

[Huaiyu-Liu]

Dear PhysIO Team,

I collected QRS and respiratory waveforms for my 7T fMRI experiment (TR = 2.1 s, 78 slices, multiband factor = 2) using Spike, and I am now using the TAPAS PhysIO toolbox to extract physiological regressors. Since PhysIO cannot read Spike files directly, I extracted the QRS and respiratory waveforms and saved them as separate .txt files for each run per participant.

However, I encountered two following issues when using the PhysIO toolbox via SPM GUI

log_scan_timing
Can log_scan_timing be left empty? I initially thought this field should contain the timing of each volume trigger (e.g., 0, 2.1, 4.2, ...). However, PhysIO only runs successfully when I leave this field empty.

onset_slice (Slice to which regressors are temporally aligned), I am not sure whether this corresponds to the reference slice used in slice-timing correction.
If so, my sequence has 78 slices with a multiband factor of 2, meaning that two sets of 39 slices are acquired simultaneously. For slice-timing correction, I used the timing of the middle slice rather than specifying slice order. In this case, should the onset_slice be 20, corresponding to the middle slice of the 39-slice group?

If not, what would be the correct value? Or I wonder is it possible to directly provide actual slice timings in the PhysIO SPM GUI?

Thank you very much in advance, and looking forward to your reply

[mrikasper]

Dear Huaiyu,

Thank you for using PhysIO.

Regarding the log_files.scan_timing parameter, this is supposed to be a filename referring to a file that stores the scan triggers and is vendor-dependent. You could also create a trigger timing file and use the 'Custom' vendor.

However, unless you have very jittered timing, it's usually sufficient to use the relative_start_acquisition parameter that defines the relative onset between scan and physiology file.

Regarding the onset slice, what you did was correct: Specify Nslices = 39 and select onset_slice = 20. There is no dedicated multiband option to specify the slice timing yet, but the general idea is as in SPM - regressors are (sub-)sampled for each slice event (Nslices).

Also have a look at this issue thread in the previous TAPAS repository, where I explained the handling of multiband data in more detail.

All the best,
Lars

[Huaiyu-Liu]

Dear Dr Kasper,

Thank you very much for your replies, very helpful.
Much appreciated!