I think we're at a point where we could actually implement this.

Currently, if a user wants to fit a bunch of scalars to dmri data, and then summarize them in bundles, they would use something like the hbcd_scalar_maps recon workflow. This does the bundle estimation, the scalar estimation and the mapping of the scalars to the bundles. This is great if I already know the exact set of scalars I want and the exact method of bundle estimation.

However, what happens often is that a new method for bundle estimation (eg @ameliecr's FOD autotrack) is introduced and we want to see how those scalars look on the new bundles. Or some new scalar estimator is added and we want to see how it looks on some bundles we've already estimated. Our group really would like to run one recon workflow that estimates lots of scalar maps, then a bunch of bundle estimators and be able to combine them.

Some things that would be helpful

• output dir can't be one of the input datasets
• qsirecon_suffix in the scalar_ingress node will specify there the output tsv will go