Hi Colin
On 11 Mar 2012, at 11:04, Colin Reveley wrote:
> Hello
>
> We are trialing some new equipment. I wonder, all parameters and data being equal, whether xfibres in 4.1.9 would be expected to process data faster than xfibres in 4.1.8 (and I mean a significant speedup like 10-50%).
>
Not really.
> secondly: one issue with tractography in general is difficulty in terminating in gray matter rather than propagating through it. Termination can be achieved by masking, but I wonder also if, in principle or practice one could constrain by one or more other tensor derived maps.
>
> It's already possibly to constrain by FA. But what about spherical/oblate/prolate maps? gray matter tends to produce spheroids with a spherical character while white does not. In the few cases where WM tensor spheroids do have a high intensity from the westin spherical measure they also tend to have a high value on the westin planar (flat tensor, indicating fibre crossing) measure. GM voxels rarely show a high value on the planar measure.
>
> using the just the spherical, or better the WSpherical, WPlanar and WLinear measures together as a constraint, I would suggest termination at the WM/GM boundary could be achieved.
>
> If ones xibres data is quite rich in the GM, one could go further than this and even terminate fairly accurately near Layer IV of selected cases of sensory cortex.
>
> basically, if one can constrain by FA, why not some other aribtrary tensor map or combination of maps?
You could use Westin's indices, don't have to stick with the FA. In FSL though we tend not to use heuristics on DTI-derived maps to terminate tracking, which is commonly done in deterministic tracking. Note that in probtrackx2 you are able to use surfaces to constrain tracking, so you can obtain directly e.g. WM/GM boundaries from anatomical scans.
Cheers
Stam
|