Dose-Volume Histogram

A dose-volume histogram (DVH) condenses the three-dimensional dose inside one structure into a single curve. It comes in two forms:

• Cumulative DVH — the form you see in the planning system. For each dose on the horizontal axis, the curve shows how much of the structure (in % or cc) receives at least that dose. It always starts at 100% at zero dose and falls as the dose rises.
• Differential DVH — a true histogram: how much of the structure receives a dose within each small dose bin. Its shape shows where most of the volume sits.

Put simply: the cumulative DVH answers "how much of this organ gets X Gy or more?"

Almost every plan-evaluation number is read off a DVH: the volume receiving a dose (Vx), the dose to a volume (such as D95% or D0.03cc), the mean dose and the maximum dose. Overlaying the DVHs of rival plans for the same patient shows at a glance how uniform each target dose is and how much dose reaches each organ at risk.

A DVH has one big blind spot: it throws away where the dose is. Drzymala and colleagues' classic description of the DVH makes the point that, because positional information is lost, it should not be the only criterion for judging a plan. A cold spot at the edge of the target and one in its center can produce the same target DVH.

1. Check which structures are plotted and that each one is the right structure — the DVH can only be as good as the contour.
2. Read the protocol's metrics from the curves, with the units the protocol uses (Gy or %, cc or %).
3. Compare plans by overlaying their DVHs for the same structures.
4. Then review the dose slice by slice, because the DVH cannot show location.

A few things change a DVH without changing the plan:

• Overlapping targets. AAPM TG-263 asks that DVH metrics for multiple dose levels state whether a lower-dose PTV excludes the higher-dose PTV (segmented) or includes it (non-segmented). A non-segmented low-dose PTV shows a "foot" of high dose where it overlaps the boost; the two give very different numbers.
• Calculation settings. Single-point values at the ends of the curve depend on the calculation grid size and placement, which is why near-maximum and near-minimum doses are often used instead.
• Structure definition. Whether a structure includes or excludes the target, or is a PRV, changes the whole curve.