WIP: Real world value mapping

[matthew-brett]

Rebased changes from : #143

Please see the comments there.

Uses Real World Value Mapping SOP class - see : ftp://medical.nema.org/medical/dicom/final/sup103_ft.pdf

These changes need tests. Also need thought about scaling of different volumes, and relationship of scaling to usual DICOM scaling with RescaleSlope and RescaleIntercept

[matthew-brett]

More research reveals more pain.

There can be many RWVM for each image.

https://groups.google.com/d/msg/comp.protocols.dicom/C6kE4X4FvLc/tWPoNvlO3DQJ

The mappings apply to ranges of stored (pixel) values, and the ranges can overlap, so more than one mapping applies to a particular pixel value.

There can be one mapping for each frame.

There's a "macro" for this as well.

http://www.dabsoft.ch/dicom/3/C.7.6.16.2.11/

Even in cases without real world value mappings, there is the the Modality LUT - http://www.dabsoft.ch/dicom/3/C.11.1/

So, this might need a little thought. But certainly we should use the RWVMs when the others aren't there, at least.

[matthew-brett]

Further investigation reasonably reveals that the example image has a RealValueWorldMappingSequence (tag 0040, 9096), with one element. This is therefore the 'Macro' and not the 'SOP'. I am not sure how the 'SOP' would get into the dicom file because the SOP would be a separate service object pair; the image should be a single service object pair, here SOP class UID = '1.2.840.10008.5.1.4.1.1.4' (MR Image Storage).

So, I think we can ignore the SOPs and just look at the Macro, in exactly the way you've done it.

I guess we can also ignore the LUT tables I mentioned above too.

So, just needs some tests. I'll put some in.