Personally I don't really think this is a very fair thing to ask. TBSS is an off the shelf tool that lots of people use. There might be ways to improve it, but honestly properly evaluating what constitutes registration improvement is a lot trickier than is widely realized. For example, there are plenty of papers and people claiming that ANTS > FNIRT; however, when we actually tested the functional/neuroanatomical alignment of the two algorithms, FNIRT clearly wins because it doesn't overfit cortical folding patterns. One also needs to characterize distortions, and it didn't immediately appear like the below reference using Elastix did that. To really improve white matter alignment, one should use fiber orientation information and still I think the skeletonization projection is the best that one is going to be able to do in cases of topological incompatibilities.
Overall, I would either make the point that you want to stay consistent with the field using TBSS, you can cite Coalson et al 2018 PNAS that shows FNIRT > ANTS if you feel it helps, and see if you can avoid having to do this.
Matt.
On 9/11/19, 3:43 PM, "FSL - FMRIB's Software Library on behalf of Joselisa Paiva" <[log in to unmask] on behalf of [log in to unmask]> wrote:
Dear colleagues,
We have recently submitted a manuscript in which DTI data was processed using TBSS. However, our paper’s reviewer seems unsure of our approach and suggested replacing the standard nonlinear registration step (FNIRT) with an alternative registration tool, as follows:
“Given that all the significant results appear to stem from the TBSS analysis, and such analyses are very sensitive to intersubject registration, I would like to see the authors compare their results with those obtained using an alternative registration program, such as Elastix (https://github.com/SuperElastix/elastix). If the same significant clusters are found using both pipelines, I would be much more confident of the results.”
We are aware that there are many non-linear registration tools available (e.g. Elastix, ANTs, SyN…) and that the feasibility and effect of incorporating the (Elastix) improved registration has been previously tested in TBSS in the paper of “de Groot et al 2013 Neuroimage”. Nevertheless, as far as I know, FNIRT is still being internally employed in the current version of TBSS and I wonder why, given the claimed superiority of Elastix (de Groot et al 2013 Neuroimage).
We also understand that the standard FSL-TBSS approach is reliable enough (especially when applied to a good quality dataset, with little head motion) and that it allows us to replicate the methodology employed in previous studies regarding the condition. We would highly appreciate if anyone could shed some light on this and on how to best answer this comment from the reviewer.
Kind regards,
Joselisa
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
________________________________
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|