[Camino-users] procstreamlines feature survey

Philip A Cook cookpa at mail.med.upenn.edu
Tue Dec 6 13:37:42 PST 2011


Hi all,

I am working on some updates to procstreamlines, to address recent concerns and to enable easier post-processing and visualization. The major changes I am working on right now are:

- Output tracts in physical space (physical space defined by the image header)

- Enable output of compressed nii.gz images

- Output images in the same image format and physical space as the input, so if you pass -seedfile someSeeds.nii.gz, you'll get output as *.nii.gz

- Allow seed ROIs to be 32-bit ints, which can be non-continuous (this will make it easier to use Freesurfer labels)


In the process of making these changes, I need to rethink some of the output options that have been in procstreamlines for a long time. I'd like to ask for some feedback about what features of procstreamlines are being used, and what use cases you'd like to support. I'm open to adding new programs as well. 

Basically, I am wondering if constructing probability maps on a per-voxel basis is useful. I usually use the "Anatomical Connectivty Map" (-outputacm), which combines all input streamlines into a single density image. It was originally implemented to test the "Anatomical Connectivity Map" idea (http://dx.doi.org/10.1016/j.neuroimage.2010.08.069) where you track over all white matter and sum up the results. 

I've found -outputacm to be the most useful option, because it allows great flexibility in processing the streamlines before combining them into an image. I almost always want to visualize connectivity over an ROI and not for a single voxel.

Aside from -outputacm, the use case I am mostly interested in is defining a connectivity matrix between different regions. conmap appears to be broken right now, but it can be fixed, or changed, if you've used conmap and have feedback I'd be especially interested in hearing from you. 

Thanks,


Phil


More information about the Camino-users mailing list