Dose and distance metrics.
The vocabulary of plan evaluation — D95, V20, mean dose, max dose — computed from the DVH for every mapped structure in every case, and available to queries, charts, scorecards, and monitoring as one consistent set of numbers.
Endpoints bind to mapped structures
A dose endpoint in Radial is always attached to a mapped structure, not a raw string. Parotid_L Dmean means the same thing across every case in the cohort because mapping settled what the parotid is first. That is why curation precedes metrics, and why cohort-level statistics are comparable at all.
Beyond dose and volume
Alongside Dx and Vx, Radial supports distance-based constraint types: overlap-volume and distance-to-target histograms (OVH and DTH), written in the same modified Mayo shorthand as the dose endpoints. That lets a protocol express geometry-aware expectations, and a reviewer see a per-patient curve behind the number.
Equivalent dose, declared
Any dose endpoint can apply equivalent-dose scaling: EQDx at a declared reference fraction size, α/β, and model, with EQD2 the common case. Declared, because the parameters are clinical: they are always visible, never invisible defaults. Targets are deliberately not converted: Radial separates them by fractionation instead of normalizing away a difference that matters.
Custom metrics: your algebra, computed live
Define any derived metric as free-form algebra over plan fields and dose endpoints bound to real structures. A live preview shows how many cohort cases actually compute and how many come back null before you save, and broken expressions are refused outright. Saved metrics become first-class: columns in the metrics tables, fields in charts, and signals monitoring can watch.