Hi  - yes that may be a source of bias.

Have you considered using FIRST to get individualised accumbens? This may be more accurate than the atlas. 

Also for the orbital mask, you should make sure to use a conservative threshold on the group mask. Although looking at it, even a liberal threshold misses substantial portions of the orbital cortex, so maybe best to find some other mask (perhaps from Connectome Workbench)? 

Cheers,

Saad

On 19 Dec 2016, at 10:35, Noelia Martinez Molina <[log in to unmask]COM> wrote:

Dear FSL experts,

I've conducted probabilistic tractography in my DTI dataset to dissect the accumbofrontal tract. To do so, I specified the nucleus accumbens (a region in the ventral striatum) as seed mask. This region of interest was obtained from the Harvard Oxford subcortical atlas in the fsl package and transformed to diffusion native space for each subject. Similarly, I included as target a termination mask of the orbitofrontal cortex taken from the Harvard Oxford cortical atlas and transformed to diffusion native space.

My concern is that in some subjects these two masks appear very close in space when I do the visual inspection in fslview and was wondering whether this proximity could somehow be biasing the connectivity values and increasing the probability of false positives (i.e. getting a tract by chance).  

Does this make any sense?

I would appreciate any help on this issue.

Thank you in advance for your time.

Best wishes,
Noelia