Allow simultaneous estimation of TE-(in)dependence metrics

[tsalo]

Closes #1224.

The only current blocker is test data- we need something EPTI-ish to test this on.

Changes proposed in this pull request:

• Add --simultaneous-dependence-metrics parameter to tedana CLI.
• Add estimate_simultaneously parameter to compute_f_maps.
  • If False, calculate TE-(in)dependence marginal F-statistics separately, as usual.
  • If True, calculate TE-(in)dependence partial F-statistics via a single model.
• Add estimate_simultaneously parameter to getfbounds.
  • If False, use dfd = n_independent_sources - 1 as before.
  • If True, use dfd = n_independent_sources - 2.

[codecov]

Codecov Report

❌ Patch coverage is 47.76119% with 35 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.54%. Comparing base (586542b) to head (704a1f8).
⚠️ Report is 29 commits behind head on main.

Files with missing lines	Patch %	Lines
tedana/metrics/dependence.py	41.66%	34 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1357      +/-   ##
==========================================
- Coverage   90.19%   89.54%   -0.65%     
==========================================
  Files          30       30              
  Lines        4893     4927      +34     
  Branches      823      826       +3     
==========================================
- Hits         4413     4412       -1     
- Misses        318      352      +34     
- Partials      162      163       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[tsalo]

@katielamar would you be willing to check this over to see if it does what you described in your 2025 ISMRM poster?

[tsalo]

@handwerkerd is there any chance you have some shareable EPTI data we could use for testing?

[katielamar]

sure. I will look this over tomorrow and confirm. As for EPTI data, I can see if we can share some of our data.

[katielamar]

@tsalo, I ran your code (using estimate_simultaneously = True for calculate_f_maps) on the 5 EPTI tapping datasets used for ISMRM. The fstat maps match mine, but the final denoised EPTI datasets are different. It looks like the getfbounds in stats.py in your code wasn't updated. So if the combined model is used, the degrees of freedom would be dfn=1, dfd = n_independent_sources -2 instead of dfn=1, dfd = n_independent_sources -1. Let me know if you want me to try running it again on my data with the update.

[tsalo]

@katielamar Thank you! I just pushed your fix.

[handwerkerd]

FWIW, an alternative to having EPITI test data, woudl be to take the 5 echo data, mask with a subset of voxels to reduce processing time, and interpolate more echoes. I think we'd need to locally write out the new files to test the full tedana workflow, but I have thoughts on how to do it without excessive computational cost.