I think there is some difference of opinion on this in the consortium. Personally, I think that the suitability of the HCP FIX training file to other scanners remains to be empirically established.
cheers, -MH -- Michael Harms, Ph.D. ----------------------------------------------------------- Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave.Tel: 314-747-6173 St. Louis, MO 63110Email: [email protected] On 8/9/16, 4:01 PM, "Glasser, Matthew" <[email protected]> wrote: If the acquisition parameters are similar I would not worry too much about differing scanners at least within a manufacturer (e.g. Siemens Trio, Skyra, Prisma, Connectom should perform similarly). Peace, Matt. On 8/9/16, 12:33 PM, "Harms, Michael" <[email protected]> wrote: > >Hi Michael, >If you use the HCP FIX training file, I would check it on some your data. >I don’t think we have a good sense yet of how well the HCP training file >generalizes to data acquired on other scanners. > >cheers, >-MH > >-- >Michael Harms, Ph.D. > >----------------------------------------------------------- >Conte Center for the Neuroscience of Mental Disorders >Washington University School of Medicine >Department of Psychiatry, Box 8134 >660 South Euclid Ave.Tel: 314-747-6173 >St. Louis, MO 63110Email: [email protected] > > > > >On 8/4/16, 7:21 PM, "[email protected] on behalf of >Glasser, Matthew" <[email protected] on behalf of >[email protected]> wrote: > >If you are acquiring HCP-style data, it would probably be fine to use >ICA+FIX based on HCP training. > >Matt. > >On 8/4/16, 9:35 AM, "[email protected] on behalf of >Burgess, Gregory" <[email protected] on behalf of >[email protected]> wrote: > >>> On Aug 4, 2016, at 8:52 AM, Michael F.W. Dreyfuss >>><[email protected]> wrote: >>> >>> Thank you, >>> >>> Is it recommended to use the motion regressors with or without >>>detrending on data that has been processed with the HCP preprocessing >>>scripts? >> >>You should perform the same filtering on your confound regressors (and >>task model) that you are performing on your data. So, I¹d start with the >>original Movement_Regressors.txt. >> >> >>> Is there a recommended cutoff for censoring time points or a >>>recommended threshold of mean relative displacement for excluding a >>>subject or run? >> >>One of the interesting difficulties with HCP data (and occasionally other >>MB fMRI data from other groups) is that motion parameters from MB data >>pick up on some source of variability that is not typically visible in >>motion parameters estimated from fMRI data with lower spatial and >>temporal resolution. So, censoring thresholds that are typical in lower >>resolution fMRI might result in censoring a considerable proportion of >>your data. You might want to look at DVARS to see if it can detect >>apparent motion artifact more cleanly. >> >>Either way, the best strategy is to look at your data to set your own >>threshold. (Look at strategies from Power, J. D., Mitra, A., Laumann, T. >>O., Snyder, A. Z., Schlaggar, B. L., & Petersen, S. E. (2014). Methods to >>detect, characterize, and remove motion artifact in resting state fMRI. >>NeuroImage, 84, 320341. http://doi.org/10.1016/j.neuroimage.2013.08.048) >> >> >>> I installed FIX and can implement it on my data, but I can only have it >>>trained to data from your scanners at Wash U since I do not have enough >>>additional data from our scanner here to train it on, so I do not feel >>>comfortable using it on my data. >> >>You could consider hand categorizing the ICA components, if you feel >>comfortable making those judgments and you don¹t have tons of data. There >>are also some other automated approaches such as AROMA that have become >>popular with folks. I have no experience with AROMA yet, but I¹d imagine >>that it would rate spike-like ICA components as noise. >> >> >>--Greg >> >>____________________________________________________________________ >>Greg Burgess, Ph.D. >>Staff Scientist, Human Connectome Project >>Washington University School of Medicine >>Department of Psychiatry >>Phone: 314-362-7864 >>Email: [email protected] >> >>> >>> Thank you, >>> Michael >>> From: Burgess, Gregory <[email protected]> >>> Sent: Tuesday, August 2, 2016 2:30:10 PM >>> To: Michael F.W. Dreyfuss >>> Cc: [email protected] >>> Subject: Re: [HCP-Users] Movement Confounds and Censoring >>> >>> > On Aug 2, 2016, at 10:17 AM, Michael F.W. Dreyfuss >>><[email protected]> wrote: >>> >>> > >>> >>> > Hi, I have some basic questions about the movement parameters that >>>are put out by the HCP preprocessing scripts. >>> >>> > >>> >>> > 1) What is the difference between Movement_Regressors_dt.txt and >>>Movement_Regressors.txt? I would like to use one as a confound. >>> >>> >>> >>> The Movement_Regressors.txt are the parameters computed from the motion >>>correction algorithm. The Movement_Regressors_dt.txt is the same file >>>put through a detrend command to remove the linear trend and mean. >>> >>> >>> >>> > >>> >>> > 2) What does each column in these represent? I'm assuming 6 >>>directions of displacement, and 3 for rotation? >>> >>> >>> >>> The 12 columns represent Displacement {x, y, z}, Rotation {x, y, z}, >>>Backward Derivative of Displacement {x, y, z}, and Backward Derivative >>>of Rotation {x, y, z} >>> >>> >>> >>> > >>> >>> > 3) What are the units in these and the other movement files in the >>>output? mm? voxels? >>> >>> >>> >>> The units are mm for displacements and degrees for rotations. >>> >>> >>> >>> > >>> >>> > 4) I would like to censor time points with too much motion in >>>Movement_AbsoluteRMS.txt. Is there a specific movement cutoff you >>>recommend? I have made a script that produces a censor file in the fsl >>>format (i.e. columns of all 0s with a 1 at the time point that is >>>censored). Is there a simple way to add that censor file into the >>>existing analysis scripts along with the motion as a confound? >>> >>> >>> >>> First, I would suggest using the Movement_RelativeRMS.txt file for >>>censoring instead. The AbsoluteRMS reflects the total movement from the >>>motion correlation registration target, which is the SBRef volume >>>collected at the beginning of the scan. In contrast, the RelativeRMS >>>reflects the amount of motion from the previous time point, which should >>>be considered a more direct measurement of movement on that specific >>>time point. >>> >>> >>> >>> I¹m not sure to which existing analysis scripts you are referring. We >>>don¹t release analyses of individual scans. So, you should probably >>>include these censoring confounds as covariates in whatever >>>preprocessing you¹re doing for the individual scans. >>> >>> >>> >>> After saying all of that, it is worth mentioning that FIX denoising: >>> >>> - Regresses a set of 24 motion parameters (six rigid body parameters, >>>their backward derivatives, and the squares of those 12 columns) >>> >>> - Regresses noise component variance (orthogonal to signal components) >>>from the scan-level ICA. >>> >>> >>> >>> In practice, the behavior of FIX is similar to regressing movement >>>paremeters and voxel-specific censoring (plus other artifact cleanup). >>>If you utilize FIX preprocessed data, you don¹t need to regress motion >>>parameters, and you might feel comfortable skipping censoring as well. >>> >>> >>> >>> --Greg >>> >>> >>> >>> ____________________________________________________________________ >>> >>> Greg Burgess, Ph.D. >>> >>> Staff Scientist, Human Connectome Project >>> >>> Washington University School of Medicine >>> >>> Department of Psychiatry >>> >>> Phone: 314-362-7864 >>> >>> Email: [email protected] >>> >>> >>> >>> >>> >>> > >>> >>> > Thank you, >>> >>> > Michael >>> >>> > _______________________________________________ >>> >>> > HCP-Users mailing list >>> >>> > [email protected] >>> >>> > >>>https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.humanconnectom >>>e >>>.org_mailman_listinfo_hcp-2Dusers&d=DQIGaQ&c=lb62iw4YL4RFalcE2hQUQealT9- >>>R >>>Xrryqt9KZX2qu2s&r=rPclmYysc_z1plf99IoNsmxWf1JolkKMmL6bXnYFSwg&m=FMg6DAhR >>>Q >>>HwDPdwDze2CIg4j1-kMkVc9h-XJ0PXmDWQ&s=ck-KAIP80cxbdHtG0fMXtiT8T_-4QzuQ-lN >>>V >>>QlPrDs8&e= >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> ________________________________ >>> >>> 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. >>> >> >> >> >>________________________________ >>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 > > >________________________________ >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 > ________________________________ 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. ________________________________ 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
