Matthew Brett

OK - time for this one to go: I'll tag its demise and leave this issue open. https://github.com/nipy/nipy/pull/313

Oops - yes - we should probably return complex here - don't you think?

@effigies - yes I think `dtype=np.float32` is a know-what-you're-doing case. At least, if they read their own code, they shouldn't be suprised they got a float32!

Oh - I was thinking only that, if the image-implied dtype was complex, we'd return complex128 by default, instead of float64, and we'd respect the explicit dtype whatver that was.

I don't know AFNI, but I am sure SPM prefers left-is-left and I believe FSL does too. So, suggested keep left-is-left as default and allow left-is-right as an option.

Sorry to be slow on this one. @csytracy - would you be interested in making this an option?

Sorry to take so long to get back to you on this one. Yes, an extra parameter to the plot_map function would be a good option - would you mind...

Hi - @csytracy - if you are interested to get this done, we could merge it soon, we will probably release in a few days.

As I remember it - [dcmstack](https://github.com/moloney/dcmstack) can read multiframe DICOMs at least - will it read your DICOMs?

Hmm - yes - totally agree - but I think this is Numpy.distutils' fault - at least - that's where all the messages seem to be coming from. I guess...