# HBN Preprocessing Meeting #1

## Attendees (alphabetical by last name)

- Matt Cieslak (Penn BBL)
- Alexandre Franco (CMI)
- John Kruper (NRDG/UW)
- Mike Milham (CMI)
- Adam Richie-Halford (NRDG/UW)
- Ariel Rokem (NRDG/UW)
- Ethan Roy (NRDG/Stanford)
- Ted Satterthwaite (Penn BBL)
- Valerie Sydnor (Penn BBL)
- Jason Yeatman (NRDG/Stanford)

## Minutes

0. Study Scope
- There was some discussion about whether or not to include Site-SI and Site-CUNY. We agreed to include every site and annotate differences.
- In particular, some subjects from Site-CBIC were scanned using the ABCD protocol. This should be noted.
- We will include Site-SI, but also note the potential harmonization issues.
- There are some participants in the FCP-INDI S3 bucket that are not listed in `participants.tsv` files. Adam will send a list of these participants to Alex and Alex will remedy.
- Penn BBL has put considerable amount of work into curating the HBN data for Site-RU and Site-CBIC. Ted will work with Alex to put that on FCP-INDI S3. They also have a todo item to repeat this curation with Site-SI and Site-CUNY, however, the timeframe for this is unknown.
0. Preprocessing parameters
- Matt will send out an example html report with all of the preprocessing parameters
- eddy linear vs quadratic discussion: quadratic is the eddy default so it's what most other groups do. Matt set linear as the default in qsiprep. We agreed to do a comparison of linear vs quadratic in a sample that overlaps with the already processed Site-RU subjects, excluding obliques.
- We agreed on 1.8mm isotropic output resolution for all subjects
- For Site-CBIC, we will use VNAV T1w

0. Prerequisite qsiprep enhancements
- Oblique scans
=> Fixed in an upcoming qsiprep release
- Desired additional output report components
=> Ribbons of WM/GM on the registration plots
- Matt and Adam are also working on dmriprep-viewer which will help with visualization of output. It should be possible to generate the required json reports for this tool using the `--reports-only` flag for qsiprep. This way we can start the preprocessing without waiting on dmriprep-viewer development.
0. QA protocol
- QA team members (so far):
- Adam Richie-Halford
- Ariel Rokem
- Valerie Sydnor
- The following qsiprep output will be useful for QA at the group level:
- Neighboring DWI correlation
- Changes in FWHM compared to Raw data
- Eddy CNR average
- T1w dice coefficient
- Head motion mean/max
- We decided to use Braindr for QA of all subjects, with five slices per volume and the following images as input:
- b=0 reference image
- eddy CNR
0. Computation
- Ariel volunteered the NRDG group to do the preprocessing using AWS resources. This may cost as much as $1.00 per subject. Everyone agreed.
0. Open dataset
- We agreed to store the generated data on FCP-INDI S3:
- Curated input data
- qsiprep container
- qsiprep output in derivatives directory
- We decided not to put the qsiprep working directory files on S3
- Alex will ask Mike about granting write access to FCP-INDI S3 to Ariel and Adam so that they can directly upload qsiprep results.
0. Publication
- Sections:
- the HBN dataset
- the need for standardized preprocessing
- qsiprep
- preprocessing at scale, challenges and solutions
- BrainDr / QC
- visualization section (e.g. dmriprep-viewer)
- Who is doing the work:
- Ariel and Adam agreed to spearhead the project and paper
- We will benefit greatly from the data curation work already completed by Penn BBL and from the development of qsiprep.
0. When to meet again
- We agreed to meet again after Site-RU and Site-CBIC have been preprocessed. We will discuss:
- BrainDr
- Any issues with Site-SI and Site-CUNY
- dmriprep-viewer
- Given the estimated due dates for the tasks below, this meeting will be after 04 Sep 2020.

## Action Items

| Todo | Assignee | Due By |
| :- | :- | :- |
| ~~Compile Todo items and send to attendees~~ | Adam Richie-Halford | 11 Aug 2020 |
| Send list of subjects not in `participants.tsv` to Alex | Adam Richie-Halford | 12 Aug 2020 |
| Update participants.tsv to include the aforesaid subjects | Alex Franco | 21 Aug 2020 |
| Upload curated data for Site-RU and Site-CBIC to FCP-INDI S3 | Ted Satterthwaite | 17 Aug 2020 |
| Repeat curation for Site-SI and Site-CUNY, upload to FCP-INDI S3 | Ted Satterthwaite | 04 Sep 2020 |
| Confirm that the curated data json sidecars for Site-CBIC will make qsiprep use the VNAV T1w | Ted Satterthwaite | 17 Aug 2020
| Send out example html report with all of the preprocessing parameters | Matt Cieslak | 14 Aug 2020 |
| Work on dmriprep-viewer | Adam Richie-Halford and Matt Cieslak | Ongoing |
| Add ribbons of WM/GM on registration plots to output reports (either in the output html or in dmriprep-viewer) | Matt Cieslak and Adam Richie-Halford | 28 Aug 2020 |
| Ask Mike about granting Ariel and Adam write access to FCP-INDI S3 bucket | Alex Franco | 14 Aug 2020 |
| Once curated data is on S3 and FCP-INDI S3 write access is obtained, run qsiprep on all subjects in Site-RU and Site-CBIC. Reconfirm qsiprep params before production run | Adam Richie-Halford | 04 Sep 2020 |
| Do it again for Site-SI and Site-CUNY | Adam Richie-Halford | 11 Sep 2020 |
| Set up braindr | Adam Richie-Halford and Ariel Rokem | 18 Sep 2020 |