Hi Michael,

Indeed I am not quite ready to start supporting the MSMAll pipeline and MSMSulc 
until we have config files and a public binary that agree with each other and 
for which we have validated that the results perform as well as in the 
publications and all the licensing stuff is sorted out.  In answer to your 
questions:

  1.  The templates will always be based on group data.  Individual subject 
versions of RSNs are made using weighted regression.  This is detailed in the 
supplemental methods of the Nature paper.
  2.  These data come from fMRI fluctations, not HCP-specific tasks.  One 
certainly can choose after the fact what modalities to use (possible options 
include C = Connectivity (RSNs), A = Architecture (myelin), T = Topography 
(visuotopic “RSNs”)).
  3.  That’s basically right.  We also run RestingStateStats because it 
produces some files we like to use in MSMAll.  I think for registration if you 
only had task it would be better to treat the task data as resting state (i.e. 
consider all of the BOLD variance rather than just the variance related to the 
task) to improve one’s coverage and CNR.

Matt.

From: 
<[email protected]<mailto:[email protected]>>
 on behalf of Emma Robinson 
<[email protected]<mailto:[email protected]>>
Date: Friday, September 2, 2016 at 8:21 AM
To: "Stevens, Michael" 
<[email protected]<mailto:[email protected]>>
Cc: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Subject: Re: [HCP-Users] MSM All

Hi Michael,

I think I should highlight that the version you have is from the 2014 
Neuroimage paper and not the one used by the HCP. I will add you to the list of 
people requiring the new version once we have finished reoptimising the 
registration parameters. This is something we are currently having to do as the 
code evolved from the version used for the multi modal parcellation.

I am privately supporting the testing of the beta version I have supplied to 
you at the moment, and am open to requests from other interested parties (under 
the disclaimer that this is beta software and we do not supply optimal 
parameters for running it). This version will be in the next FSL release which 
will be released very soon. At that point my support (for use of the software) 
will go to the FSL list. The current version of the code will run with the HCP 
pipelines and data, but the alignment will not be quite as as good in some 
places. I also have a new version of the code to distribute so I will send this 
in due course. Note also when available the HCP parcellation version will not 
immediately go into FSL, but will be available either on request or though the 
HCP.

As for the rest of your questions, they are more orientated to the pipeline 
itself, which Matt would be better placed to answer. However, I have tried 
running ICA+FIX on task data incidentally and it works.

I hope this helps.

Emma

On 2 September 2016 at 13:45, Stevens, Michael 
<[email protected]<mailto:[email protected]>> wrote:
Hi Tim and Matt,

Thanks again for such a great series of lectures earlier this week at the HCP 
workshop.  The course got me motivated to try to implement the MSMAll approach 
on data we’ve produced here locally.  I’ve had the binary for MSMAll since 
earlier this year, but haven’t yet tried to put the pieces together.  I wasn’t 
sure if you wanted to start publicly supporting the MSMAll stuff.  But if it’s 
OK to ask questions about MSMAll script implementation, I have two simple 
script usage questions and a pipeline logic question:


1)      Is the logic of MSMAll that it should be run on single subjects, using 
resting state-derived templates, weights, and myelin maps from each individual 
dataset?  Alternatively, should each individual subject be run using 
group-derived templates, weights, and myelin maps?  E.g., instead of 
‘rfMRI_REST_Atlas_MSMAll_2_d41_WRN_DeDrift_hp2000_clean_PCA.ica-dREPLACED_ROW_vn/melogic_oIC.dscalar.nii’,
 do we find the corresponding file from our own output for each subject we run 
and reset the variables within the MSMAll script to fit?

2)      If individual data is supposed to be used, can the MSMAll script be run 
without TopographicRegressor and TopographicMap variables?  The bash code looks 
like it currently aborts if these are missing.  My understanding was that these 
Togographic data came from specific HCP-produced data that I wouldn’t 
necessarily have in my own experiments.  But if I’m wrong, let me know.

3)      To incorporate MSMAll properly for our own data, would the logic of 
overall pipeline use be:

a.      Run all the FreeSurfer scripts and fMRIVolume script as usual

b.      Run the fMRISurface script for REST (using MSMSulc settings)

c.      Run the HCP ICA+FIX script for REST

d.      Use the info from all that output to feed MSMAll for each subject – 
which then specifies the OTHER fMRI task timeseries to process (and would you 
put the REST timeseries *back* into the processing at this point as well so 
it’s ALSO mapped using MSMAll?  Or would that confuse the script and/or output?)


At that point, would it be done?  Matt, you suggested at the HCP course that 
running ICA+FIX on the task-based fMRI data wouldn’t hurt (something I’d 
strongly agree with is worth trying).  But other than that, would any further 
steps be needed after MSMAll before doing fMRI stats?

Again, if you’re not yet supporting this, no worries.  I’ll wait until you are 
and ask again later.  But it’d be cool to see this work on the HCP-compliant 
data we’ve been collecting for a few of my R01-funded projects.

Thanks,
Mike

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]<mailto:[email protected]>
http://lists.humanconnectome.org/mailman/listinfo/hcp-users


_______________________________________________
HCP-Users mailing list
[email protected]<mailto:[email protected]>
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

________________________________
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.

_______________________________________________
HCP-Users mailing list
[email protected]
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

Reply via email to