Philipp,
I think we'll need to take a look at the data with this one. Not clear
otherwise exactly what has gone wrong. So, please could you tar up the
group feat (gfeat) directory, put it on a web or ftp site, and email me
the address.
Thanks.
Cheers, Mark.
Mark Woolrich.
Oxford University Centre for Functional MRI of the Brain (FMRIB),
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
Work: +44-(0)-1865-222713, Mobile: +44-(0)-7808-727745
Email: [log in to unmask]
On Fri, 10 Oct 2003, Philipp Prahs wrote:
> Dear FSLers,
>
> I have analyzed a series of scans from a single person during pain
> stimulation using fsl 3.1 first level analysis. The scans have been repeated
> 6 times. I am now trying to get an average statistic for these scans. I set
> up the design matrix in FEAT 5.1 higher level analysis with one EV with ones
> for each of the six scans to get the mean activation.
> As posted earlier on this list I checked the registration and it works
> pretty fine. Anyhow, when FLAME is run I get lots of "ndtri domain error"s
> while the percentage increases. Switching to "Simple OLS" or FLAME stage 1
> does not help, but fewer ndtri domain errors are produced.
> As far as I can tell there are no other error messages as in earlier
> postings. Also FEAT does produce a nice looking result.
> I have no idea what a ndtri domain error is or what could cause it.
> My question is now whether the produced results are correct and I could just
> ignore those errors (like if the errors are from voxels from the margins of
> the scans where not all the scans are overlapping) or are these errors
> affecting the statistics?
>
> Thanks very much for any help.
>
> Philipp Prahs
> Dept. of Radiology
> Ludwig-Maximilians-University Munich
> Mail: [log in to unmask]
>
|