[MINC-development] MINC and ITK

[sean at rogue-research.com]

> While a discussion on which itk interface to support could be a worthwhile
> one, I'm rather surprised that it could hinge on the fact that Leila, one
> of the two main developers behind MINC 2, is somehow not deemed a "core
> MINC developer." So let's argue on technical merit rather than the
> suggestion that the ITK interface could not be supported by Leila or the
> rest of our group in Toronto.

Apologies if my words gave offense, none was intended. I was not aware of
her involvement.  I was not trying diminish (or inflate) anyone's
contributions, competencies, or otherwise. Indeed, I am only interested in
technical matters. (Notice that I cc'ed Leila on my original post!)

My interest is merely in ITK being able to open any medical image file
format.

ITK currently does DICOM, Analyze, and NIfTI. It also does MINC2 (thanks
Leila!) and PAR/REC, but these last two are in ITK's 'Review' directory,
which I should have described in my first post: it acts like the 'entry
point' for code contributed to ITK. It is not meant as a final place for
such code. It is not included in default builds. It is not built and
tested nightly (by anyone except myself, as far as I can tell).

Unlike (I gather) most people on this list, I am not part of any MNI /
McGill / MINC circle, I have no knowledge of your inner workings, I don't
know who the developers are. I simply googled for info on MINC and ITK
integration and found stuff like this on the MINC wiki:

http://en.wikibooks.org/wiki/MINC/EZMINC/ITK_Integration
"If you are not satisfied with MINC2 only interface provided by ITK, or
need access to MINC file attributes, or want to read/write 4D data use ITK
interface provided by EZMINC"

http://en.wikibooks.org/wiki/MINC/Future/Tconf20110223
"ITK & VTK – currently with limited support."

My impression therefore was that the "core MINC developers" (whoever you
all are!) acknowledge that there is work to be done in this area. I only
wanted to inform this list of the discussion that started here:

http://www.itk.org/pipermail/insight-users/2011-May/040953.html

As Andrew suggests, I think "one 'sanctioned' interface" would be best, so
that ITK can include it. The one ITK currently has is great, but your own
wiki implies that it does not support: MINC1, 4D data, and MINC file
attributes. I believe it also does not support .xfm files. Support for all
those would be great since the idea here is for ITK to open any image
format you throw at it. It could well help adoption of MINC; ITK, and thus
all apps based on it, can already open all the 'competitor' formats.

>I was of the impression that Vlad was volunteering to perhaps
>integrate his code or ideas into the current ITK MINC reader so that
>MINC 1 file input was possible?  or has this already been done?

In ITK itself anyway, itkMINC2ImageIO hasn't changed since 2008:

http://itk.org/gitweb?p=ITK.git;a=history;f=Code/Review/itkMINC2ImageIO.cxx;h=32006d7f597eff40629e76c7d3266872e9c248dc;hb=f338f8a0ce5fe94e8a02cd3e7e9c2a728e9c1a00

I'm sure both codebases have their pros and cons (I have not really
looked), but an integration combining the best of each sounds ideal! If
such a thing is to happen, now is probably the best time (from the
perspective of ITK), since they are cleaning through their Review folder. 
I know next to nothing of MINC, but know something of ITK, and could
contribute a little time to help in this endeavour.

> think, I can be persuaded to be official minc-ITK supporter - how do
> I sign up?

Leila and I are already members, if anyone else is interested, I
suggesting joining the ITK mailing list(s):
http://www.itk.org/mailman/listinfo/insight-users
http://www.itk.org/mailman/listinfo/insight-developers

Cheers,

Sean