Choosing how hot the target gets
A long tail on the target DVH is a decision, not an accident. The prescription isodose line sets it and the ring constraints hold it.
Built in·updated 2026-08-14
Two plans can both cover the PTV and look nothing alike on the DVH. One drops almost vertically at prescription; the other carries a long tail out past 130%. Neither is a mistake — they are different answers, and which one you get is decided by two settings before the optimizer runs.
The prescription isodose line decides the heterogeneity. Prescribe to the 80% line and the hottest voxel in the target sits near 1/0.80 = 125% of prescription. Prescribe to 65% and it sits near 154%. Nothing else has to change: the same objectives on the same case produce a different DVH shape purely because you moved the line the dose is normalised to.
The ring constraints decide whether it stays inside the target. A hot core is only useful if the dose falls off immediately outside the PTV. Left alone, the optimizer will happily let the same hot region spill. Nested rings with stepped upper limits are what stop it.
The published setup
Wang et al. built the geometry explicitly — one structure inside the target and three outside, all carved out of a partial body limited to PTV ±5 cm superiorly and inferiorly:
| Structure | Definition |
|---|---|
| inner | PTV minus the ITV margin |
| 5 mm ring | PartialBody − (PTV + 5 mm) |
| 8 mm ring | PartialBody − (PTV + 8 mm) |
| 2 cm ring | PartialBody − (PTV + 2 cm) |
Upper limits across them ranged from 30% to 110% of prescription at priority 150 — the ladder is what makes it work, not any one number: the near ring holds a value close to prescription, and each ring further out holds a lower one, so the constraint set describes the fall-off you want rather than a single wall.
With that geometry and a P-IDL of 75–80% (their average optimum was 77.9 ± 0.9%), they measured R50 down 29.1%, conformity index up 10.3%, V20 down 26.9%, V5 down 13.9% and mean lung dose down 16.7% against the delivered clinical plans. Dose concentration in the target core rose 23.7% — the tail is where the sparing came from.
Which way to go
Toward heterogeneous — lower the prescription line, release the target max objective, and tighten the rings. You buy gradient, low-dose spillage and treatment time. You lose homogeneity index, and the distribution becomes harder to compare against another plan.
Toward homogeneous — raise the prescription line toward 100%, constrain target max, and let the rings sit looser. You buy an interpretable distribution and an HI that scores well. You pay in gradient, and usually in monitor units.
Where the literature disagrees, and why it matters
This is not a settled number. The same platform's prescription isodose skill reports an optimum of 59–69% for lung on a conventional linac, against 75–80% here. Both are real measurements; they optimised different things. Read what a study minimised before borrowing its line — a paper minimising mean lung dose and a paper minimising gradient index will not agree, and neither is wrong.
On a scored case this is a scoring decision as much as a clinical one. Homogeneity index and gradient index pull in opposite directions. If the criteria weight HI heavily, a heterogeneous plan is being penalised for the thing that produced its fall-off. Read the point weights before you choose the line.
Where the numbers come from
- Optimal prescription isodose line in SBRT for lung tumor treatment with volumetric-modulated arc therapy
Journal of Radiosurgery and SBRT · 2020
Optimal P-IDL 75-80% (mean 77.9 +/- 0.9%). With nested rings at 5 mm, 8 mm and 2 cm carved from a partial body, upper limits 30-110% of prescription at priority 150: R50 -29.1%, CI +10.3%, V20 -26.9%, V5 -13.9%, mean lung dose -16.7%, target core dose concentration +23.7%.
How to do it in
- Plan → Goals → Prescription
- Fx, Dose, and the normalisation VOI Normalising to a target Min is the homogeneous choice; a lower prescription isodose is what creates the hot core.
- Target Goals → Goal Type
- Min / Max per VOI Drop or loosen the target Max to go heterogeneous; set it near prescription with a high weight to go homogeneous.
- Target Goals → Weight
- e.g. 200 Relative, not absolute. Raise the one you are trading for, do not raise everything.
- Critical Goals
- Max per ring structure Stepped down ring by ring — near ring close to prescription, outer rings lower.
- NTO tab
- On (BODY) Reference-point alternative to explicit ring structures.
- Dose Statistics Table
- CI · nCI · GI · HI columns HI and GI move in opposite directions; this is where you see what the choice cost.
Everything this skill describes lives on one screen: Plan → Goals.
Set the line in `Prescription`. Fx and Dose are the obvious fields; the one that decides heterogeneity is the normalisation beside them — the plan header reads Min of target_bone in the capture above, meaning the prescription is normalised so that structure's minimum sits at 100 Gy. Normalising to a target minimum is the homogeneous choice. Normalising so a lower isodose carries the prescription is what produces the hot core.
Add the goals per structure. Each VOI takes a Min row, a Max row, or both, with a Weight and a Specified Dose. Toward heterogeneous, leave the target Max loose or drop it; toward homogeneous, add it close to prescription with a high weight — the capture shows Max 100.50 against a 100 Gy prescription, which is the tight end.
Rings go in `Critical Goals`, or use `NTO` instead. The Critical Goals table below takes the same Min/Max form for the ring structures. The NTO tab does the same job with reference points rather than structures, and on this platform is often the faster route — see the shell-ladder skill for the comparison.
Read the result in the Dose Statistics Table.
HI near 1.0 means the target dose is flat; a heterogeneous plan pushes it up and buys a lower GI in exchange. Those two columns are the trade this skill is about, and they sit next to each other.
Captures are from a phantom QA plan, cropped to the controls.
Link to one system:Accuray Precision