I agree with Mike and Gordon. Our best guess right now is that the motion correction registration target (i.e., SBRef) is corrupted, and that is affecting the efficacy of motion correction. Previous testing in healthy normal adults suggested that the SBRef was the best motion correction target. But, we’ve seen that the SBRef can be corrupted by motion during the SBRef collection (immediately prior to the MB scan volumes) in high motion individuals (e.g., kids, clinical populations). In those cases, it’s possible that a different motion correction target would yield better results.
The mcflirt_acc.sh script in the HCP pipelines allows for alternate “Scout” images (like the mean image or mid-volume), or in the absence of a motion correction reference image the scripts will create a reference image from the average of the first 10 frames. Even with a good motion correction target, large motions may not be completely corrected. From your image, it appears that all of the other frames were adequately corrected. There was no absolute displacement that was uncorrected. Only the timepoints of large displacement were inadequately corrected. To salvage data of this sort, you should seriously consider using frame censoring (aka “scrubbing”) or motion outlier regressors (a la fsl_motion_outliers) in your subsequent analysis. --Greg ____________________________________________________________________ Greg Burgess, Ph.D. Staff Scientist, Human Connectome Project Washington University School of Medicine Department of Anatomy and Neurobiology Phone: 314-362-7864 Email: [email protected] On Oct 24, 2014, at 1:10 PM, Xu, Junqian <[email protected]> wrote: > Greg, > > We’ve observed very similar issue in our Sinai non-compliant subject data. I > briefly discussed the issue with Greg Burgess at WU, which didn’t lead to a > conclusive solution. I think it may be worthwhile for us to put the > troublesome motion correction data together and cross-check our examination > methods to get to the bottom of this issue. > > Gordon > >> On Oct 24, 2014, at 9:36 AM, Book, Gregory <[email protected]> >> wrote: >> >> I processed some of our fMRI own data through the HCP pipelines and now I’m >> looking at the AnalysisDir/MNINonLinear/Results/taskname/taskname.nii.gz >> file. I’m assuming this is the 4D file which should be motion corrected, >> undistorted, and skull stripped. And that this file should be able to be >> used directly in stats processing without any further steps? >> >> However, I still see significant motion when viewing the 4D file in a cine >> view. The brain mask is constant, but the image under the mask moves. Did >> the mcflirt step not work, or is this normal? >> >> The animated .gif can be seen here: http://olinnidb.org/preview.gif The >> major motion occurs around volumes 150 and 200-250. >> -G >> >> _________________________________________________ >> Gregory Book >> Senior Technology Manager >> Olin Neuropsychiatry Research Center, Institute of Living, Hartford Hospital >> 200 Retreat Avenue >> Hartford, CT 06106 >> Tel: 860-545-7267 Fax: 860-545-7797 >> [email protected] >> http://nidb.sourceforge.net >> >> >> This e-mail message, including any attachments, is for the sole use of the >> intended recipient(s) and may contain confidential and privileged >> information. Any unauthorized review, use, disclosure, or distribution is >> prohibited. If you are not the intended recipient, or an employee or agent >> responsible for delivering the message to the intended recipient, please >> contact the sender by reply e-mail and destroy all copies of the original >> message, including any attachments. >> _______________________________________________ >> HCP-Users mailing list >> [email protected] >> http://lists.humanconnectome.org/mailman/listinfo/hcp-users >> > > _______________________________________________ > HCP-Users mailing list > [email protected] > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
