rsfMRI alternates oppositely-encoded volumes


I am new to fMRI processing and have run into an interesting issue with my rsfMRI volumes: every other 3D volume is flipped in the A-P direction (around the y-axis) and I think this is how they were encoded in the scan, because the distortion direction also switches with each volume.

Does anyone have any suggestions for how to go about resolving this? Do you know of any fMRI-specific tools that might deal with this?

I considered flipping the volumes myself and using FSL TOPUP to distortion correct but wasn’t sure whether there were other implications to the flipping that needed to be accounted for.

Any help is more than welcome!

1 Like

Oh wow - I never heard about such sequence. Is it raw data? Which scanner/sequence? Is this sequence described somewhere in detail? If this is something more common we need to figure out how to adjust BIDS.

One thing I can think of is to split the data into two fake runs each with different phase encoding directions, flip one of them so they would be consistent, put the data through FMRIPREP and then reconstruct the original order by interleaving the volumes from the two runs (they will be aligned and should be distortion corrected).

Hi Chris! Thanks for the reply.

It is raw data, from several scanners (Siemens TrioTim, GE Signa + Discovery, Philips Achieva) from the PING (Pediatric Imaging Neurocognition and Genetics) Study. Their protocols for each scanner type are linked here: but there may be some discrepancies between what is written there and the actual DICOM (as even within scanners, the images are from multiple sites).

I’ll try what you suggested! I also have “calibration” scans for some of the volumes – to help with distortion correction I’m assuming. Would there be a good way to integrate those (whether that’s FSL TOPUP or otherwise)?

Thanks again!

This is quite surprising. I looked at Siemens protocol. Just to clarify you are saying that the ep2d_bold_A-P file has every other volume in opposite phase encoding (so volume 1 is A-P, volume 2 is P-A etc.)?

The scans in particular I was looking at are from a GE (which I only just realised may be the only ones with this issue) but yes, it is as you described (1st A-P, 2nd P-A). I went looking for a Siemens scan and it does not have this problem but I only checked one out of a few hundred.