Hi David,
Thank you for your reply – this is very helpful!
Jess.
On 2017-05-16, 2:05 PM, "FSL - FMRIB's Software Library on behalf of David Roalf" <[log in to unmask] on behalf of [log in to unmask]> wrote:
Hi Jess,
If at all possible I would highly recommend going beyond simple visual inspection. We published a manuscript in Neuroimage in 2016 (see citation below) indicating four metrics that you can compute from the raw diffusion data that will give you a good sense as to whether your DWI data is good, okay or ugly. This includes measures of motion, temporal signal to noise, and two measures of voxel intensity on a volume by volume basis (essentially looking for 'intensity spikes'). This was done in over 1200 diffusion weighted data sets, which included manual inspection of each and every volume. We have compiled this as a script, which is available here: http://davidroalf.com/script_download/
To run the script it requires that AFNI and FSL are both installed and in your PATH environment. Once run it produces a simple text file for the data that includes the QA metrics. The motion metric is currently restricted to using interspersed b=0 images, but as you will see from the manuscript each of the four measures listed above does an excellent job of capturing poor data. We have not systematically use the volume by volume voxel intensity measures, but they do correspond to detectable motion artifact and this metric could be used to identify problematic volumes
As Mahmoud indicated the new eddy tool does give out a nice text file of the motion and eddy values (but this does mean that you have to run eddy in order to get you QA data, which depending on the number of volumes and the speed of your computer/cluster can take a while).
As we mention in the paper, there are several other QA tools (DTI Prep; RESTORE: TORTOISE) than can be used to assess and address some of the artifact.
Our Neuroimage paper: Roalf DR, Quarmley M, Elliott MA, Satterthwaite TD, Vandekar SN, Ruparel K, Gennatas ED, Calkins ME, Moore TM, Hopson R, Prabhakaran K, Jackson CT, Verma R, Hakonarson H, Gur RC, Gur RE.Neuroimage. 2016 Jan 15;125:903-19. doi: 10.1016/j.neuroimage.2015.10.068. Epub 2015 Oct 28
Good luck.
~David
>>>>>>>>
As far as I know just eye-balling. Any method you apply to images would be some sort of preprocessing ...
On Tue, May 16, 2017 at 12:07 PM, Jessica Reynolds <[log in to unmask]> wrote:
Thanks Mahmoud.
Do you know if there is a way I am able to check this motion before undertaking any other form of preprocessing?
Thanks,
Jess.
From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Mahmoud <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Tuesday, May 16, 2017 at 9:51 AM
To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] Identifying motion corrupted volumes in DTI .nii files
The new version of eddy function will give you output text files which identify the volumes that eddy considered as outliers. There is more info about it on FSL wiki.
Good luck,
Mahmoud
On Tue, May 16, 2017 at 11:35 AM, SUBSCRIBE FSL Jess <[log in to unmask]> wrote:
I am new to FSL and am wondering whether there is a way in FSL to identify which volumes in a raw DTI .nii file have too much motion/are poor quality and should be excluded. I will also be visually checking over scans, and using motion correction, but am working with a paediatric sample so will need to exclude some volumes completely. If I am unable to do this in FSL, is anyone able to recommend an alternative way or program to identify these volumes?
Thanks in advance,
Jess.
|