arueter1
commented
2 years ago @LuciMoore I addressed all of the above issues (to be released by Barry next week) except for the following: 2.2. Input Data Subsets Breakdown: • possibly make separate section for data structure
2.7 Field Map Selection Process • 1st sentence: replace with link to BIDS definition of pair of pepolar fieldmaps
3.3 Pipeline Stages • stage 10: add link to cifti-conn-wrapper repo --> not planning on adressin this since we will be only using BICEPS moving forward
Would be good to address this in a further ABCC review when the new version is pushed.
LuciMoore
Overall notes: • make it so the table of contents on the left lists subsections (similar to the table of contents shown on the right side of the page) • perhaps switch order of Inputs and Pipelines section • use A, B, C, etc instead of numbers for subsection labels. or 1.1, 1.2 etc • perhaps explain the overall purpose of this documentation and how it can be used
Table of Contents: • ABCC Documentation Summary: should cite ABCC paper and DAIC here instead of just having the useful links section • How to Contribute: • should we remove Amanda's email and just ask people to post GitHub issues? or maybe use the DCAN lab email? • move this section to the end • Documentation Guide - remove this section and change settings so that it opens new tab when you click on links. remove this section from all of the subsections • Table of Contents: • move this section to under ABCC Doc Summary • rename as "Overview"
Release Notes: • Possibly move both 1.1 ABCC Collection and 1.2 Background to Table of Contents section since it doesn't seem to be directly related to releases. And then the first section will be "Ongoing Releases" • move this Release Notes section to further down, perhaps between Derivatives and Useful Links somewhere • 1.2 Background: need to state explicitly that this data is being housed on NDA
Inputs General comments about this section: • a lot of the info could go into a section named "Metadata" (such as 5. MRI Acquisition Parameters) • add information about sidecar jsons and other information required to run pipeline? Or at least include link to pipeline RTD that describes how to run pipeline • it should be more clear that this section is describing how the data was processed, not instructions on how you would process the data yourself
2.1. About this Document: • move description about pipeline processing to the "Pipeline" section - doesn't seem to be relevant to information here. can keep last 2 sentences • add both task and rest fMRI data • state clearly that ABCC includes BIDS-formatted raw data (NIFTIS) and processed data and this section describes what raw data was used and how it was processed (as opposed to describing how one would process it themselves - many people in group are getting confused on this point)
2.2. Input Data Subsets Breakdown: • possibly use or * instead of #
• delete 1st sentence "Section 3 .." because this is confusing - unclear which Section 3 it is referring to
• describe what is meant by "input data subsets" since this has a specific meaning according to NDA
• inputs.anat. - should the file name include "_rec_normalized" in "sub-#/ses-#/anat/sub-#_ses-#(_rec-normalized)(run-#)(T1w|T2w).json"
• possibly make separate section for data structure
2.3 DAIC Quality Control • consider moving this information to beginning of RTDs • move ABCD repo info to 2.4 DICOM to BIDS conversion section • go to NDA open workshops to gain more information to add to this section 2.6 Scanner Differences • 2nd paragraph: should specify that ANTS was used for all anatomical data preprocessing • any description of pipeline should be in pipeline description, confusing to include that information here • the info currently in this section only discusses ANTs - we should either flesh out this section with more scanner differences, change the name, or make it more complete. it also reads as somewhat anecdotal • the info in this section could also be moved to 2.5 MRI acquisition parameters 2.7 Field Map Selection Process • 1st sentence: replace with link to BIDS definition of pair of pepolar fieldmaps • combine 2.7 with section 2.8 (BIDS Field Map "IntendedFor" Metadata) • specify that pair of fmaps were selected for all scans, currently language is a bit unclear 2.9 Diffusion-weighted Imaging (DWI) • should allude to fact that we are currently starting to process the DWI data and so processed dwi data will be included in future release • middle paragraph: should say these bval and bvec files were provided to us by the DAIC due to formatting irregularities that come with the files in DICOMs 2.10 BIDS Modality-agnostic files • replace "small blurb" with more formal language • typo in word "dataset"
Pipeline 3.1 About this Doc • remove "1" from other pipeline sentence, confusing 3.2 Pipeline Overview • link goes to ABCD-STUDY repo - should also link to whichever version of pipeline used to do processing • update link to DockerHub to our new UMN DockerHub • add section about which version of the pipeline was used for processing (looks like it's 1.2.0 from what it says in 1.1 Release Notes) or just link back to Release Notes section • Zenodo: explicitly state version, or include link to all versions • Bigger picture: need to figure out which data subsets were processed with which version and make that explicit including specific release links, especially for future releases 3.3 Pipeline Stages • part 3 motion censoring: temporal mask range is actually 0 to 1.0 mm now • could be more explicit that stages 8-10 (custom clean, filemapper, and connectivity matrix generation) are not technically a part of the ABCD-HCP pipeline • stage 10: add link to cifti-conn-wrapper repo
Derivatives 4.1 About this Document • fMRIPrep and QSIPrep should be added elsewhere and not mentioned here- just add to useful links section? • specify that these are the derivatives that are available on the NDA, not the full pipeline outputs 4.2/4.3 Anatomical/Functional • perhaps present this information in a table format • figure out different want to insert blanks (currently using ) - perhaps use pipe notation instead, and add note at bottom that clarifies how to interpret this • specify that all the parcellated time series are concatenated • "Prerequisites for each fMRI type to have 8 derivative data subset:" reword to make it more clear and move to Inputs section 4.4 Derivative Data Subsets Breakdown • change "Section 3 and onward" to "This section" - in general, perhaps remove this language so that it sounds less like a paper • put this section into a table format • consider just linking to FreeSurfer documentation for some of this information (https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/AnatomicalROI#IndividualStatsfiles) • parcellations: can just list parcellations used, don't need to have so much redundancy - could also put this info into a table as well (data subset name, derivatives, file path) with description above or below • Motion Framewise displacements: change FD threshold range to 0-1.0 mm 4.5 Executive Summary • add link to executive summary documentation 4.8 Caveats • perhaps add this information to release notes or some place that's updated more regularly • double check that executive summaries are still not available yet for anat-only subjects and update text as needed • clarify what the last point on exec summaries is referring to ("labeled as fMRI scan_type in records")
5 Recommendations General note: add more context for what each specific recommendation actually is. the information throughout this section is frequently presented without context
5.1 About this Document • explain more about what these recommendations are for, generally need more context 5.2 The BIDS Participant Files and Matched Groups • add more context for why we are showing table • explain that we recommend that when using data for analyses, you should split the subjects following this table that we've provided so that the demographics are similar between groups 5.3 BIDS Quality Control File • add context for specific recommendation - that one should consider using these QC metrics to determine which data to include in their analyses 5.4 Downloading and Unpacking Data • put this section before "BIDS participant files and matched groups" section 5.7 DCAN Labs Software • move this section from recommendations entirely and merge with customclean and filemapper sections under Pipeline section • also, for custom clean, should remove info about GUI since we don't use that anymore 5.8 BIDS Folder Layout • move this section to Derivatives section, perhaps at the top? • and remember to update links within the section as needed