[Insight-users] DTI data.. can't read

[Gordon Kindlmann]

hello,

> Never mind; my fault. The data now reads fine with ITK. The "mask"  
> component is not respected or preserved though, is it?

No, the mask component is removed (nrrdCrop()'ed away) as part of the  
read (Insight/Code/IO/itkNrrdImageIO.cxx, line 700 or so).

I saw no clean way of making that mask field available, except to  
return it when some asks for a 3-D scalar field from that file, but  
that seemed too clever for its own good, and besides most people  
would probably expect that they'd be getting the B0 (non-diffusion- 
weighted) image, which in this case would be false.

To me the notion of accompanying the data with a mask always seemed  
fundamental, since the tensor values are constructed by what is  
really a division between MR signals, and if you ended up doing  
something close to 0/0, you want to record that somewhere close by so  
that people don't bank on the results.  Anyway.

Of course, if you engineer a better solution for reading in the mask,  
or keeping it around in computations, that would be great.

Gordon