[Mrtrix-discussion] connectivity p and distance from seed

[Luis Concha]

Hello.

Since tracking errors build up as streamlines propagate from the seed, and
uncertainties in the PDFs are added, it is inevitable that the probability
of connection of a given voxel from a seed is dependent upon its distance.
As such, voxels far away from the seed often have very low probabilities of
connection to the seed (even <1%). Compensating these low probabilities is
often performed by multiplying the visitation counts per voxel by the
average euclidean distance from the seed (this is how fsl does
it<https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0805&L=FSL&D=0&P=80972>).
The result, however, is uninterpretable, as it is impossible to disentangle
the contribution of distance from diffusion-derived connectivity.

Does anybody in this list have a strategy to obtain reasonably high
probabilities of connectivity to a distant seed that can be compared to
those obtained in voxels that are closer to the seed? Could you recommend
some papers discussing this issue? Any help will be most appreciated.

Thanks.

Luis Concha.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20121004/169bfa95/attachment.html