Hi Jasper,
If you are interested in movement in a certain direction, you have to check one of the first two plots (mm threshold only works for translation, rotation you have to specify in radians), If you just want to have an indication of the overall movement regardless of which type of, then you should look at the third one.
The absolute movement is certainly important to consider, but the relative (from one timepoint to the next) is more critical I think, so relative displacement is what you should check. That said, I think it is probably a little too conservative to exclude the subject altogether if there is a strong motion. You could also try looking into scrubbing the volume at which that motion occurred. Jeanette Mumford has a great video on her youtube channel in which she explains the rationale and the method of censoring motion outliers. Maybe you should have a look.
I think one of these two:
https://www.youtube.com/watch?v=BgaLqLHgIes
https://www.youtube.com/watch?v=YTpEKjUdMG8
Best,
Eduard
________________________________________
From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of J. van Oort [[log in to unmask]]
Sent: 21 May 2018 13:24
To: [log in to unmask]
Subject: [Spam] [FSL] exclusion based on sudden movement in feat report log
Dear all,
At the moment I am preprocessing my MRI data with FSL. We would like to exclude subjects with a sudden movement in the translation direction > 1 voxel (2 mm). We would like to do this based on the outcome of the Feat. I have learned from a colleague that I should look at the second plot of the pre-stats of the feat report log (MCFLIRT estimated translations (mm)). In the attachment I have added a file for which I thought the sudden movement was a little less than 2 mm. Now another colleague told me I have to look at the third plot of feat report log (MCFLIRT estimated mean displacement (mm)) and that I have to look at the line for relative movement in that plot.
Now I have the following questions:
- At which plot should I look?
- If I have to look at the second plot: is it only important to look how large the movement of one of the lines is and if there is a sudden increase or decrease > 2 mm? Or is it also important how large the movement is with respect to the point where the line started at time point 0 (at the start of scanning)? And is it also important if the line crosses the x-axis (the 0.0 line)?
I hope someone can help me with this.
Best, Jasper
|