How to do DKI in qsiprep?

It seems like DKI should be possible with a typical multishell acquisition, yet there is no reconstruction is qsiprep that produces the kurtosis scalars. Why is that? What am I missing?

Thank you,

-Dianne

I do believe it was added to qsiprep in Add DKI, DKI-Micro, MSDKI and SMT2 from DIPY · Issue #206 · PennLINC/qsiprep · GitHub. Possibly not exposed in the user facing api? Or not sufficiently well-documented?

One way you can already get it is by running the pyafq recon pipeline and changing this line of the input json to:

"scalars": "['dki_fa', 'dki_md']",

which would then call DKI through pyAFQ

Hi @Dianne_Patterson,

If you need whole brain DKI maps, use the dipy_dki recon spec. See the specification here: https://github.com/PennLINC/qsiprep/blob/master/qsiprep/data/pipelines/dipy_dki.json

Best,
Steven

Thank you! I did not realize dki was already available. I will try it.

@Ariel_Rokem thank you for pointing this out! I played around a little with adding scalars to the py_afq spec (I guessed that rd might work, and it did). I will try these dki scalars as well.

Is there guidance somewhere about which scalars are available and how they are named? I feel like I’m blind to the parameters available in these JSON files.

Sorry to be slow here. The guidance about the metrics available through pyAFQ is on this page: The pyAFQ API methods — AFQ 1.1 documentation, but let us know if the documentation could be improved in any way and we’ll do our best.

Thank you! That looks like what I was hoping for!