ECAT file support

[idoimaging]

Currently, EcatImage is not included in all_image_classes in imageclasses.py. Adding it and calling load() on an ECAT file gives an error in ecat.py line 562:

        i = iter(affs)
        first = i.next()

Changing the second line to the Python 3 syntax first = next(i) seems to address this issue, but there may be other Python 2 code in the EcatImage class.

[effigies]

There are some open issues for adding EcatImage to nibabel.load: #95 / BIAP7

I believe the current way to load is:

import nibabel
nibabel.ecat.load(fname)

So it looks like the issue here is that we're never testing this code. I wonder if you have a multi-frame ecat of a phantom that we could add to our test battery.

I'll open a pull request with your suggested fix.

[effigies]

Actually, since you found this issue, would you care to submit a pull request?

[idoimaging]

OK, I'll get on to both of those. Also apologies for not first finding the earlier issue #95.

[idoimaging]

I have plenty of multi-frame Ecat phantom images but they are all 27 MB per frame. Here's one: https://www.dropbox.com/s/nh3d2vyijkf69td/ecat_03fr.v?dl=0

[effigies]

Thanks. We may be able to reduce the size a bit one way or another. If not, we have some repositories of large test files we can add it to.

[effigies]

@idoimaging Do you have any opposition to releasing that image to the public domain (e.g., by the CC0 license)? Or perhaps some other license? I'm considering adding it to nipy-ecattest, which currently has one file licensed under CC0.

I can also post an image that is identical, except with a data block of all zeros.

[idoimaging]

That's fine, though it's not a phantom obviously, and is huge. If you don't need the voxels, they could certainly go.

[effigies]

Okay, I'll do the zeroed out version. If there's any metadata I should destroy let me know.

[effigies]

It turns out editing ECAT files is actually quite difficult. I suppose I'll go with the original file.