week 06/26 eric

[ebridge2]

by Thurs:

• add multithreading support multithreading stack 1
  • multithreading tutorial
  • good example stack example 2
• use AFNI and add anatomical preprocessing leg to improve low-intensity white-matter performance: ~25 failures -> 10 failures out of 460
• single-application of registration transforms (no multiplications until end): runtime goes from 23 -> 12 mins, and 2xfewer rounds
• update to add friction 24 parameter: Discr .756 -> .795
• update nuisance to do compcor and use priors for segmentation: Discr .795 -> .88
• edges instead of full-brains when looking at atlases

Thurs -> Monday (some goals will be done by Thursday):
most deliverables are in group presentation.

• Code
  • add more comments throughout code
  • test anatomical preprocessing at 1mm, and check whether it is worthwhile from a registration quality/computational perspective anat 1mm specs
  • merge in greg's recent changes: tried but greg ripped out the fMRI stuff which overwrote my commits, so this is non-trivial unfortunately. will rotate back at a later time.
  • pep8-ify
  • better bids sweeper implementation
• Motion-statistics/QA for functional prep
  • plotly -> matplotlib motion parameters
  • motion plots as pngs instead of html for consistency
  • use FD according to power et. al.
  • verify that group motion qa works still
• updates to templates/documentation
  • redo entire schematic to reflect new workflow
  • update tutorial
  • add documentation for what is happening at each step somewhere (to serve as "pseudo appendix for manuscript")
• Results
  • analyze CPAC results
  • finish running FNGS datasets
  • analyze FNGS results
• Slides
  • update pseudo code
  • update slides with better content for Greg (Tuesday)
  • make slides good for presenting to group
• Manuscript
  • think of 5 figures/captions we could show that would make an interesting manuscript (see below)
• misc:
  • email pedro with a link to the docker, a link to a tutorial, and a link to the schematic
  • submit application for JHU job
  • submit application for gigantum job
• grant:
  • submit AWS grant

[ebridge2]

FNGS_figures.pdf