Contouring practice

Contour it yourself.
Then see how you did.

Delineate organs at risk and target volumes on real planning CT, slice by slice, with the tools you would use in a planning system. Submit, and your contour is compared with the expert structure set voxel by voxel.

The gap it fills

Contouring is learned by doing,
and checked by comparison.

Students rarely get to contour a case and find out, precisely, where they went wrong. RT Contour uses the expert structure sets already attached to each dataset as the reference, so feedback is immediate, quantitative and specific to the slice.

Real cases, real anatomy

Every case is a genuine planning CT with its clinical structure set. The same datasets power the image-matching trainers, so a case you have matched is a case you can contour.

Scored against the expert

Dice similarity, 95% Hausdorff distance in millimetres, volume difference and slice coverage — the same measures used in contouring quality-assurance studies.

Guidance where you need it

Each structure carries anatomy notes, boundary definitions and the pitfalls that catch people out, with a link to the relevant eviQ protocol — read it while you contour, not afterwards.

How it works

Four steps, one honest number.

Pick a case

Choose a dataset and the structure you want to practise — bladder, rectum, parotid, brainstem, a PTV, and more.

Contour it

Pencil and brush with adaptive density limits, magic wand, eraser, window/level, ruler and HU probe, across axial, coronal and sagittal views.

Submit

Your outline is closed and filled, then compared with the expert structure voxel by voxel in the same coordinate space.

See the difference

A colour-coded overlay shows agreement, over-contouring and missed regions, with plain-language feedback and a score out of 100.

Dice similarity
Overlap between your contour and the expert on the slices you drew. Above 0.8 is usually considered clinically acceptable.
95% Hausdorff
The worst-case boundary deviation in millimetres, ignoring the extreme 5% so a single stray voxel does not dominate.
Volume difference
Whether you are systematically over- or under-contouring, as a percentage of the expert volume.
Slice coverage
How much of the structure's true extent you attempted — so three excellent slices of a forty-slice organ cannot score full marks.
The toolset

Built to feel like a planning system.

Nothing here is a novelty. Each tool exists because it is one you will actually use at a workstation.

Adaptive brushSamples the density where you start and stays inside that structure, so a stroke begun in a rib fills the rib and stops at the cortex.
Pencil with auto-closeDraw an outline; it closes the loop and fills the enclosed area on release, because expert structures are solid regions.
Magic wandClick inside a tissue to trace the connected region in one go, then tidy the edges by hand.
Three linked viewsAxial, coronal and sagittal side by side, sharing one contour. Swap which plane is large; draw in any of them.
Window / levelSoft tissue, bone, lung and brain presets plus manual level and width in Hounsfield units.
Ruler and HU probeMeasure distances in millimetres and sample tissue density before you commit to a boundary.

A note on guidelines. The teaching notes in RT Contour are written in-house for this platform. Where a published protocol governs a structure we link out to it — eviQ and other guideline bodies retain copyright over their own material. Always follow your department's protocol: this is a training tool, not a clinical one.

Ready to contour?

Sign in with the same account you use for the image-matching trainers. Your attempts are saved so you can watch your Dice score climb.