Support for complex multi-echo fMRI

[tsalo]

Summary

Functional phase data (from magnitude+phase reconstruction of fMRI data) is not currently supported as a multi-echo data type. I know that there are plans for a more robust form of support for multi-volume sequences in #346, but in the meantime I think that _phase needs to be added to the list of supported multi-echo data types.

Platform details:

Choose one:

• Local environment: OSX Mojave, Python 3.6.2

• Container

• Heudiconv version: latest, as of 5cfb6f2

[yarikoptic]

Do you have some shareable (phantom, ideally lower res) sample dicoms?

[tsalo]

I do! It's phantom data, although not particularly low resolution.

Here is the link to the data on Google Drive.

The data have six echoes, a multi band factor of 4, and were acquired on a Siemens Prisma:

• 612_Estimation_task1_6ET_MB4_SBRef: The single-band reference image (unnecessary but shared anyway).
• 613_Estimation_task1_6ET_MB4: The magnitude data.
• 614_Estimation_task1_6ET_MB4: The phase data.

[yarikoptic]

any objections if I reshare them from http://datasets.datalad.org/?dir=/dicoms (a new salo/ subdataset there). There is already http://datasets.datalad.org/?dir=/dicoms/velasco/MEEPI which we use for some rudimentary testing

[tsalo]

No objections. Thank you!

[tsalo]

@yarikoptic, should I wait for a larger refactor (e.g., the plans in #346) or would it be okay for me to open a pull request with my proposed change here?

[mgxd]

@tsalo I'd say go for it - the refactor won't come for a bit.

[yarikoptic]

Here is the link to the data on Google Drive

sorry for the delay. I've requested access now