Using BIDS with Siemens HYDI Data

Hi All,

I’ve recently acquired some HYDI data off of a Siemens scanner, and I am unsure of the naming format for this kind of data? Specifically because the raw folders are all very similarly named, so I’m not sure how to differentiate using the dwi format for naming. The raw folders are named:
I’m assuming that the field maps would be named according to the field name structure, but again, I’m unaware if the current structure accounts for ways to differentiate these four field maps from each other. I’m also confused as to why SBRef is included as I thought HYDI was a multiband method.

Any insight with this would be greatly appreciated!

Multi-band sequences acquire multiple 2D slices from a single 3D volume at the same time. This allows a lot of observations (fMRI) and allows one to collect more diffusion directions (DWI) in a given amount of scan time. So the main benefit is a very short repetition time (TR). An intrinsic consequence of a short TR is poorer signal to noise and contrast (e.g. little time for T1 recovery). Another issue with multi-band is aliasing artifacts, in particular when only a few channels are used.

Therefore, a nice feature of the CMRR sequences is that they can acquire a single-band reference image that is identical to the multi-band but acquiring each slice sequentially, leading to a longer TR and avoiding the aliasing effects.

Tools like fmriprep can leverage these single-band images.

Likewise, your sequence has images with both anterior->posterior (AP) and posterior->anterior (PA) phase encoding polarity. Both directions have EPI readout distortion, but these are in the opposite direction. Therefore, FSL’s TOPUP can combine these images to make a less distorted image. This is particularly useful for multi-band, as we are using the parallel coils for accelerating between slices rather than within slices (iPAT/SENSE/GRAPPA). In-plane acceleration like iPAT reduces spatial distortion (at the expense of SNR). Therefore, multi-band images often have substantial spatial distortion that we correct with post processing using the reversed phase encoding.

On the other hand, there is no reason to keep the “TRACE” image from this sequence. This is a derived image from the others. You can turn this off from the console’s Diffusion tab. If you really want a TRACE image, you can make a much better one after you have processed your multi band images (Denoise, DeGibbs, Eddy, TOPUP).

1 Like

Thanks so much for this Chris, and my apologies for the delay in seeing and responding to you. Your explanation on the different images was really helpful. In terms of converting to BIDS, would each of these outside of the TRACE need to be converted for preprocessing or processing? If so, are you aware of what the naming conventions that would be?