Youšll want to focus your analyses on voxels and fibers whose f>=0.05.
Also, youšll want to be sure you arenšt looking at voxels around the rim
of the brain, so it would help to use a grey and white matter segmentation
instead of a simple brain mask.
Išd also have a close read of the Behrens et al methods papers (2003,
2007) to make sure you understand what this data is.
Peace,
Matt.
On 3/29/15, 11:56 AM, "eric f" <[log in to unmask]> wrote:
>Hi, I'm a newbie to FSL and imaging analysis, so apologize in advance
>for what might be a naive question.
>
>I've been trying to understand the output data from bedpostx and am
>getting confused. I've made some scatterplots of merged_theta1 vs
>merged_phi1 and also merged_theta2 vs merged_phi2 where I use nibabel
>to read the files then convert them into numpy arrays using get_data
>and np.array. I notice that many the theta2's are extremely large, so
>I used remainder(th<i>,pi) to bring them all into the standard range.
>
>When I plot random voxels (chosen from mask value =1) about half of
>the time I get 2 clusters, one for the first fiber and one for the
>second -- which is what I expected. However, the other half of the
>time, I get phi2's scattered all over the place. When I look closer
>at these examples, the values of the theta2's and phi2's are gigantic
>--around 10^10 or so.
>
>Can someone tell me what I'm doing wrong?
>
>thanks,
>Eric
>
>
>
>--
>http://www.icsi.berkeley.edu/~ejf/
|