Adding elbow & variance information to the reports

[handwerkerd]

Summary

There are a few additional bits of information that would be useful to add to the reports. I'll list a few here.

Additional Detail

• Add elbow thresholds to the sorted kappa & rho plots so that people can see which components fell above & below the elbows. Also may how when the elbow calculations were a bit odd compared to other datasets in a study. I'm not sure the elbow criteria are currently being recorded anywhere
• Add total variance explained by the PCA/ICA as well as the total variance explained by the Accepted, Ignored, & Rejected components. I've found these numbers to be really useful to highlight a weird subject in a data set. All these values are in a combination of the tedana_[date].tsv log and can be summed from the values in ica_metrics.tsv

Next Steps

• Decide were we want to store numbers that aren't linked to a specific component. In the original me-ica, there was a header for the ICA component metrics table that included this information, but we got rid of that at some point. Should we make a tsv or json file that can currently contain a few summary values and can be expanded to include more?
• Store the elbow info somewhere and add a few markers to the reports
• Calculate the variance summations and identify where in the tedana html report we want to display that info
• While working on this, any other summary numbers we should store & display?

[handwerkerd]

The elbow information will be added when #756 is merged, so that part of this issue will be resolved. Adding the variance info to the reports would require a few more changes.

[tsalo]

If we want to report variance explained by PCA and ICA, we should also report variance explained by global signal (when used).

We can write out a run-level TSV for info like this. I'd go with that over a JSON because it should be easier to aggregate over runs/subjects. XCP-D does something similar with desc-linc_qc.tsv (which has QC metrics for each run).