Hi Christina,
it definitely may! I.e. the nasty artifact can affect any of these stages.
The signal loss of the artifact in left/right direction propagates into FA values by imitating diffusion along x (and thus commissural fibres).
However, the good news is that dtifit can "regress out" the artifact when you specify a proper confound file (the --cni flag). See Dan Gallichan's clever HBM paper for the details.
A final note (to save my life;): the artifact has been observed at scanners from other vendors than Siemens. But at least, Siemens has finally come to offer a hardware fix (which in fact seems to work).
Cheers-
Andreas
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von Christina [[log in to unmask]]
Gesendet: Dienstag, 27. Oktober 2009 17:23
An: [log in to unmask]
Betreff: [FSL] analyzing DTI data with Siemens vibration artifact
Hi I have a dataset that has been affected by the infamous Siemens vibration
artifact.
I am desperately trying to make use of the dataset and doing TBSS analysis
by following the directions which results in no significance in my areas of
interest (or anywhere).
Since the artifact happens in an area that does not include my regions of
interest I have been going in the direction of trying to mask out the
artifact part of the brain during the randomise command with the -m option.
Unsuccessfully.
At this point I am wondering if the artifact is affecting my normalization
during the tbss_2_reg step or possibly the skeletonisation and creation of
the mean FA image during the tbss_3_postreg step?
Does anyone have a suggestion?
Any direction would be appreciated!
Thanks
Chrissy
|