a) It may give the same answer, but if for nothing else than speed of execution, TaskfMRILevel1.sh should be smarter and have an option to not apply a temporal filter a second time.
b) If it turns out that the space-spanned by the FIX noise components overlaps with the task GLM (for whatever reason), then noise will be re-introduced to some degree. That’s simply the math. You’re arguing that as a practical matter that won’t be the case, or that the effect will be tiny, which may indeed turn out to be the case, but that should be empirically demonstrated first. Cheers, -MH -- Michael Harms, Ph.D. ----------------------------------------------------------- Associate Professor of Psychiatry Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave. Tel: 314-747-6173 St. Louis, MO 63110 Email: mha...@wustl.edu From: "Glasser, Matthew" <glass...@wustl.edu> Date: Tuesday, February 26, 2019 at 8:19 PM To: NEUROSCIENCE tim <tsc...@mst.edu>, Leonardo Tozzi <lto...@stanford.edu> Cc: "Harms, Michael" <mha...@wustl.edu>, "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: Re: [HCP-Users] ICA FIX output missing Regarding task analysis, I don’t really agree with Mike: a) The sICA+FIX filter is essentially equivalent to a linear detrend whereas the task filter is more aggressive. Thus, applying both is going to give essentially the same answer as applying the more aggressive one. b) Because the noise component removal is done as a non-aggressive regression (where the neural signal is protected by the neural components) and the task design is noiseless, I don’t see how the task design will reintroduce noise. There could be statistical efficiency benefits to exactly replicating the regression on the task design, but these are likely to be small. Overall, this is not an effect that should lead one to avoid cleaning data before running a task analysis, as the statistical benefits of cleaning are clear, both from the perspective of removing biases and reducing uncorrelated variance. Matt. From: Timothy Coalson <tsc...@mst.edu<mailto:tsc...@mst.edu>> Date: Tuesday, February 26, 2019 at 6:37 PM To: Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>> Cc: "Harms, Michael" <mha...@wustl.edu<mailto:mha...@wustl.edu>>, Matt Glasser <glass...@wustl.edu<mailto:glass...@wustl.edu>>, "hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>" <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Subject: Re: [HCP-Users] ICA FIX output missing That is saying that you don't have the matlab gifti library installed (or it isn't on your matlab path). Tim On Tue, Feb 26, 2019 at 6:09 PM Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>> wrote: Dear Michael, Thank you very much for all the consideration on the use of FIX for the task data. I have tried the addition you suggest. I think the command is detected, but I get the following error in tfMRI_EMOTION_RL_hp2000.ica/.fix.log: {^HUndefined function or variable 'gifti'. Error in ciftiopen (line 31) cifti = gifti([tmpfile '.gii']); Error in fix_3_clean (line 46) BO=ciftiopen('Atlas.dtseries.nii',WBC); }^H Would you have any thoughts on this? Thank you, Leonardo Tozzi, MD, PhD Williams PanLab | Postdoctoral Fellow Stanford University | 401 Quarry Rd lto...@stanford.edu<mailto:lto...@stanford.edu> | (650) 5615738 From: "Harms, Michael" <mha...@wustl.edu<mailto:mha...@wustl.edu>> Date: Tuesday, February 26, 2019 at 7:37 AM To: "Glasser, Matthew" <glass...@wustl.edu<mailto:glass...@wustl.edu>>, Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>>, "hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>" <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Cc: "Burgess, Gregory" <gburg...@wustl.edu<mailto:gburg...@wustl.edu>> Subject: Re: [HCP-Users] ICA FIX output missing Hi Leonardo, Couple things: 1) In the context of FIX, things get a little convoluted, since the FIX distribution has its own settings.sh file that needs to be set appropriately. If you’ve hard-coded the FSL_FIX_WBC variable in that settings.sh file, then the location to wb_command in the Examples/Scripts/SetUpHCPPipeline.sh isn’t necessarily relevant. In the settings.sh file for FIX on our cluster, we use the following construction: if [ -x "$(command -v wb_command)" ]; then FSL_FIX_WBC=$(command -v wb_command) else echo "ERROR in $0: wb_command (Workbench) must be in your path" exit 1 fi so that FIX does actually respect that location of wb_command that is already in your path. 2) Regarding MR-FIX and the TaskfMRIAnalysis scripts, while they may run after MR-FIX, there are two issues that need to be addressed yet: a) The temporal filter, which was presumably already applied during MR-FIX, gets applied again with TaskfMRILevel1.sh. This script needs to be modified to be smarter regarding the temporal filtering (i.e., provide an option to NOT reapply the temporal filter). b) The space spanned by the noise regressors from FIX is not regressed out of the task regressor prior to the GLM, which means that variance removed during FIX can be reintroduced during the task GLM fitting (depending on the extent to which the space spanned by the noise regressors overlaps with the task GLM). Cheers, -MH -- Michael Harms, Ph.D. ----------------------------------------------------------- Associate Professor of Psychiatry Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave. Tel: 314-747-6173 St. Louis, MO 63110 Email: mha...@wustl.edu<mailto:mha...@wustl.edu> From: "Glasser, Matthew" <glass...@wustl.edu<mailto:glass...@wustl.edu>> Date: Monday, February 25, 2019 at 6:53 PM To: Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>>, "Harms, Michael" <mha...@wustl.edu<mailto:mha...@wustl.edu>>, "hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>" <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Subject: Re: [HCP-Users] ICA FIX output missing You’ll want to be using wb_command 1.3.2. I am not aware of any modifications that are necessary to use the TaskfMRIAnalysis scripts on MR+FIX data and have analyzed hundreds of subjects after MR+FIX. As for this issue, is wb_command set properly here: https://github.com/Washington-University/HCPpipelines/blob/master/Examples/Scripts/SetUpHCPPipeline.sh What about on your ${PATH}? As for MR+FIX itself, we are only waiting on an FSL 6.0.1 release as testing has concluded successfully. Matt. From: <hcp-users-boun...@humanconnectome.org<mailto:hcp-users-boun...@humanconnectome.org>> on behalf of Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>> Date: Monday, February 25, 2019 at 5:09 PM To: "Harms, Michael" <mha...@wustl.edu<mailto:mha...@wustl.edu>>, "hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>" <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Subject: Re: [HCP-Users] ICA FIX output missing Dear Michael, Thank you for pointing me to the logfiles. It seems like the script is not finding the directory where wb_command is. In my case, I am loading it as a module in my HPC cluster. I have also put its path in ICAFIX/fix1.067/settings.sh as follows: # Set this to the location of the HCP Workbench command for your platform FSL_FIX_WBC='/share/software/user/open/workbench/1.3.1/bin/wb_command'; However, the script does not seem to “see” this path and instead uses the setting I was using on my local machine. In the logfile tfMRI_EMOTION_RL_hp2000.ica/.fix.log, I get the following error: /bin/bash: /Applications/workbench/bin_macosx64/wb_command: No such file or directory Is there another place in the scripts that is overriding my settings.sh? Concerning the length or the runs, I will look into the multirun implementation, but indeed my intention was of using the TaskfMRIAnalysis scripts to get my “activations”. Thank you, Leonardo Tozzi, MD, PhD Williams PanLab | Postdoctoral Fellow Stanford University | 401 Quarry Rd lto...@stanford.edu<mailto:lto...@stanford.edu> | (650) 5615738 From: "Harms, Michael" <mha...@wustl.edu<mailto:mha...@wustl.edu>> Date: Monday, February 25, 2019 at 9:59 AM To: Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>>, "hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>" <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Subject: Re: [HCP-Users] ICA FIX output missing Hi, The log files for ICA FIX are a bit scattered. In the .ica directory check: .fix_2b_predict.log (from the prediction – i.e,. R code) and .fix.log (from the cleaning stage). And in the .ica/fix directory, check logMatlab.txt (I believe that is from the feature extraction stage). Note that our recommendation is to use “multi-run” FIX on the task data, due to its shorter run length. We hope to have an announcement on that in the near future. In that regard, you would implement your desired filtering as part of the MR-FIX cleaning (and there is a new “polynomial detrend” option, for faster execution), although I believe that we haven’t quite gotten around to adapting the TaskfMRIAnalysis scripts to work on data from MR-FIX. Cheers, -MH -- Michael Harms, Ph.D. ----------------------------------------------------------- Associate Professor of Psychiatry Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave. Tel: 314-747-6173 St. Louis, MO 63110 Email: mha...@wustl.edu<mailto:mha...@wustl.edu> From: <hcp-users-boun...@humanconnectome.org<mailto:hcp-users-boun...@humanconnectome.org>> on behalf of Leonardo Tozzi <lto...@stanford.edu<mailto:lto...@stanford.edu>> Date: Monday, February 25, 2019 at 11:42 AM To: "hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>" <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Subject: [HCP-Users] ICA FIX output missing Dear Experts, I have been trying to use ICA FIX to denoise task data of a large number of subjects from the HYA release. I have managed to make it run with no errors on our local HPC cluster, but I still have one problem. In the vast majority of subjects, even if FIX works and produces a file (“fix4melview_HCP_hp2000_thr10.txt”) which shows which components are noise, I don’t get the final output, for example “tfMRI_EMOTION_RL_Atlas_hp2000_clean.dtseries.nii”. This is especially puzzling since it does seem to work for a minority of subjects. The matlab log also shows no errors. A related question I would have is what filter you would recommend for task data. My intention is to use a GLM on the cleaned data, so is 2000 (linear detrending) ok, since then a lower high-pass will be applied in the GLM step? Thank you very much, Leonardo Tozzi, MD, PhD Williams PanLab | Postdoctoral Fellow Stanford University | 401 Quarry Rd lto...@stanford.edu<mailto:lto...@stanford.edu> | (650) 5615738 _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org<mailto:HCP-Users@humanconnectome.org> 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 HCP-Users@humanconnectome.org<mailto:HCP-Users@humanconnectome.org> 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 HCP-Users@humanconnectome.org<mailto:HCP-Users@humanconnectome.org> 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 HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
