medinria INR files

[arokem]

Hello! Someone on the DIPY mailing list asked about INR medInria files (here). Has there ever been a previous discussion about implementation of I/O for these files? I couldn't find anything here or on the mailing list.

It looks like the Matlab files linked on this thread would be a good starting point for anyone interested in implementing this.

[arokem]

I added the prio-low tag, because I believe unless someone shows up really gung-ho to do the work, I don't think that this will get done, but others can feel free to change that.

[effigies]

I don't know of any previous discussion of this format, but if someone has a need and the time, happy to review a PR.

[anm-ranjan]

The INR file from MEDINRIA basically maps the coefficient of the diffusion tensor to the voxel co-ordinates of streamlines. If one is aware of the streamline_mapping from dipy.tracking.vox2track module one would need something like the following :

from dipy.tracking.vox2track import streamline_mapping
streamMap = streamline_mapping(streams,niftiImage.affine, True)
INR_Mask = np.zeros(niftiDimension[0], niftiDimension[1], niftiDimension[2])
for eachVoxel in streamMap :
                INR_Mask [eachVoxel[0],eachVoxel[1],eachVoxel[2]] = T

where T could be the (9,) vector or (6,) vector containing the diffusion coefficients at the given voxel. I am just not sure how to extract this T-vector.