That is a more difficult question. If the denoising you are adding doesn't need spatial information, and is being applied after FIX, etc, one possibility that might be less painful is to take the existing CIFTI data and run the denoising on that (using -cifti-convert if you need it in a different file format strictly for IO purposes). Otherwise, you may need to mess with the arguments to the pipeline scripts in order to patch in your volumes at the right location.
Tim On Mon, Nov 14, 2016 at 4:17 PM, Guy Hwang <[email protected]> wrote: > Hi Tim, > > > Thank you so much for the prompt reply. So actually I have both volume > preprocessed and surface preprocessed data using HCP v3.4.0 scripts. I took > the resting images from the volume preprocessed dataset because I wanted to > apply some denoising, regressing first before pushing them through > cifti-parcellate. Would you rather recommend I take the surface > preprocessed data and apply denoising there? > > > Thank you, > > > Guy > ------------------------------ > *From:* Timothy Coalson <[email protected]> > *Sent:* Monday, November 14, 2016 3:47:42 PM > *To:* Guy Hwang > *Cc:* [email protected] > *Subject:* Re: [HCP-Users] Question about -cifti-create-dense-timeseries > > To use the resulting dtseries.nii file with HCP data, you will first need > to map the cortical data to surfaces. The various -cifti-create commands > do not do this mapping, they are simply data conversion. You also need to > nonlinear register to MNI (for subcortical signal). The <label-volume> is > specifically to define the structures to represent as voxels in the > resulting dtseries file, which is why it says this about <label-volume> > lower in the help: > > The label volume should have some of the > label names from this list, all other label names will be ignored: > > CORTEX_LEFT > CORTEX_RIGHT > CEREBELLUM > ACCUMBENS_LEFT > ACCUMBENS_RIGHT > ... > > While that shows that it is possible to represent cortex as voxels in the > cifti format, note that it is not recommended, and that it is not what HCP > cifti files do (so you will have a very hard time using the HCP > parcellation unless you map your subject timeseries data to the surface). > The label volume for subcortical signal we use for HCP data exists in the > pipelines repository, the Atlas_ROIs.2.nii.gz file here: > > https://github.com/Washington-University/Pipelines/tree/ > master/global/templates/91282_Greyordinates > > The full process we use is fairly involved, and the HCP Pipelines ( > https://github.com/Washington-University/Pipelines) are the easiest way > to do it, when you have data matching our acquisition recommendations. > Even if your acquisitions don't entirely meet our recommendations, many of > the pipeline scripts can still be used to make the process easier. > > Tim > > > On Mon, Nov 14, 2016 at 3:11 PM, Guy Hwang <[email protected]> wrote: > >> Hello, >> >> >> I am trying to turn nifti files into dtseries.nii so that I can later use >> -cifti-parcellate to create resting state connectivity matrices. >> >> >> 1) What is <label-volume>? As in the syntax "wb_command >> -cifti-create-dense-timeseries {output}.dtseries.nii -volume >> {input}.nii.gz <label-volume>" >> >> >> 2) Where can I find the most up-to-date dtseries.nii file from HCP to use >> for the parcellation? >> >> >> I would appreciate your help. >> >> >> Thank you, >> >> >> Guy >> >> _______________________________________________ >> 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
