View source for FBIRN:AHM2006EABnotes

EAB response:

Stephen:

# much discussion of data collection and processing pipelines
# number of traveling subjects: need at least 10, rather than 6. Staggering it might be good--six on the east coast, then...
# site-specific data: 15-20 at each site would let us to an effect size assumption to let us estimate better the number of traveling subjects to have. (Possibly for a 2nd round of traveling subjects later in the project.)
# Collect as rich a dataset as possible within fMRI--especially physiological info about heartrate and breathing. The EAB says those are not optional. Cardiac noise adds to the noise in the data. GSR also a possibility and of course the breathhold. All sites should try to collect everything, but only on the traveling healthy controls (who are patient and able to lie in the scanner for a long period of time, etc.).
# Standardized FIPS pipeline including QA techniques should be made up. Tom's diagnostic toolbox could play a role.
# Scanning protocols and human factors as a source of between-group variability (e.g. head motion in Sz different from controls? Except Greg B found no group differences in stimulus-correlated movement in Sz vs controls).
## have to remain vigilant about this, including training for new personnel who join the team
## evaluation of movement while subject is on site, preferably still in the scanner. If this is uploaded to the HID, then we can see if sites drift and become careless about head movement
# NIfTI and FSL: DFWG will be working on the issues raised but it's mostly the FSL guys who need to be convinced. We are moving toward using the NIfTI format and they are happy to work with us as much as possible.
# Neuroinformatics: SRB and HID discussion, upload script fallibility, all to be worked on.

Bill: Worked with the NI group for the past two days. This is a transition time for FBIRN--we now have a lot of distributed data, preliminary tools, etc.

# Product
## Establish some identity. A package of data, software, etc. that is identified with us. (like the NA-MIC kit)
## Phase I data will be a valuable resource for the community; it's currently available but needs to be made available at a level where people can use it.
## another challenge is getting the word out to people. Other than SFN, where else? If you give a talk, give a plug for the FBIRN and make sure people know the data are available.
### E.g., to run a contest on the Phase I data. HBM's contest offered $10K as first prize (with an objective test about the data). The fMRI DC has expressed interest in a contest. Deposit at the fMRI DC as a contest dataset?
## download page: there is a cut-and-paste credit text for the data and that's good. Citations in the acknowledgments with grant number in it is important.
# Process
## The NI group is prioritizing that data be robust (backed up), that the SRB is robust, data mirror'd, etc.
## we need to look to regression testing over the coming year--we have to know whether even a pixel is changing due to an upgrade in software, etc. We need a process in place for that, similar to what the ITK group does.
# People
## there's an atmosphere of people enjoying working together, and that needs to be encouraged and supported. A programmer's week is a valuable resource to both training and building community/sharing the workload.
## Sharing the workload: We're now demanding a lot from BIRN-CC; we need to work with them closely to solve our problems.
# This is overall very encouraging. We have customers, funding, a dataset, etc. Don't underestimate the power of the NI group; they are a great resource.

Other points:

* The scanning checklist is not being uploaded and it has information that needs to be shared.
* Head movement: training for technicians regarding head positioning, etc. Make sure we're standardizing how we apply our cognitive tasks, etc.
* Gary has head movement scripts for real-time feedback; can we apply those across platforms?
* Prove Ed Bullmore wrong about correlated head-movements by group.
* ASL cognitive task tested together in a small subset of sites.
* Make a list for BIRN-CC of our top priority needs.
* Standardize the terminology on pre-processing steps and what they mean.
* Notes also coming from Cam Carter and Biff Bunney.