Using TOPUP on the fMRI data from the HCP

Hi Neurostars users.

A colleague and I are working with the HCP fMRI data (in native space) and looking to do distortion correction on it using TOPUP/FUGE.

We have the LR/RL files we need, but we’re having some difficulty estimating the total readout time from the protocol located here:

According to the formula: Readout time = (Number of echos - 1) * echo spacing, it should be something along the lines of (72 -1) * 0.58 = 41.18 should it not? This does not seem correct to me.

Further, does the multiband factor of 8 alter any of factors at all since they are simultaneous acquisitions?

Any help or suggestions you would be willing to provide would be greatly appreciated.

Thanks in advance.


You should rather look at the phase dimension which is 90 in that case.
The results would then be: (90-1)*0.58= 51,62 ms for the total readout time (FSL definition) for that fmri sequence.
The multiband factor allows a drastic reduction of TR, as only (72/8)=9 slices groups are acquired, instead of 72 if no multiband was used: The TR is shortened by a factor of 8! (slightly less since the pulse excitation needs to be a bit longer for a multiband excitation pulse).

1 Like

For reference, I like to refer to this figure:

Found in the documentation of BrainVoyager: figure 1.18

1 Like

That manual should REALLY be on the FSL documentation. It’s just so intuitive.

Thanks so much @jsein!

That said though, should 51.62 ms be reduced by a factor of 8 given the multiband or is that just what it is?

No, the multiband factor has nothing to do with the susceptibility induced distorsions.
What can play a big role though is the parallel acceleration (i.e GRAPPA, SENSE) where you sample only a factor of your ky lines and hence you decrease you total readout time by this factor. The consequence being, you get this reduction factor for your distorsion!
For example, on Siemens scanner, when you use GRAPPA =2 , you get twice less distorsions and your TotalReadoutTime is divided by 2 when compared to no GRAPPA images.
BUT, as often, no free lunch with MRI, GRAPPA images get reduced SNR by sqrt(GRAPPA factor) and are also much more sensitive to motions.

Got it.

Thanks again!