Group-level metric collection

[tsalo]

One useful application of the analysis_level group option would be to collect metrics from each run into a single TSV file.
Here are some that might be worth collecting:

1. Number of components.
2. Number of accepted components.
3. Number of rejected components.
4. Number of components for each rationale.
5. Mean GM T2*.
6. Mean WM T2*.
7. Mean CSF T2*.
8. Mean GM S0.
9. Mean WM S0.
10. Mean CSF S0.

[handwerkerd]

11. Variance explained for all components
12. Variance explained for accepted components
13. Variance explained for rejected components
14. Counts of components with each classification_tag
15. Mean GM TSNR (detrended mean/ detrended stdev) for pre-tedana, optimally combined, & post tedana
    • Note: TSNR will always increase when variance is removed, but how much it increases might be a good quality check
16. Mean WM TSNR (detrended mean/ detrended stdev) for pre-tedana, optimally combined, & post tedana
17. Mean CSF TSNR (detrended mean/ detrended stdev) for pre-tedana, optimally combined, & post tedana

[tsalo]

Great ideas!