Mapping and the audit loop.

Mapping isn't a one-time cleanup; it's a loop that compounds. Review what the dictionary can't yet decide, let every confirmation teach it, and re-run the audit to catch what surfaces later.

The review queue

Structures the dictionary can't confirm on its own are queued for review, each showing the proposed TG-263 name, the confidence, and the evidence, name and geometry both. The queue is built for speed: one keystroke per decision, with the contour in view when you need to look.

Confirm once, backfill the archive

When you confirm a local variant, Radial offers to backfill every existing structure that matches it, so a decision about one case quietly resolves the same question across years of archive. The next import of that variant maps itself.

The audit re-runs in a click

An empty queue doesn't mean a finished job. Re-run the structure audit and it re-derives its findings from the live database (e.g., unmapped structures, below-confidence survivors, and naming drift as planners and protocols change), so what surfaced since last time is in front of you, not hiding in the archive. The loop is the discipline: the dataset stays curated, not just once-curated.

Why it gets easier

The dictionary is your institution's, and it learns locally. The tenth month of mapping is faster than the first because most of your vocabulary has been confirmed once already. The work shifts from naming structures to catching exceptions.