RT Structure Set: keyhole technique for joining contours

[Віталій Омельченко]

Hi DICOM experts!

I'm currently implementing an export of RT Structure Set DICOM files. I have found that the description of the contour representation is not sufficient enough and leaves more open questions than answers. I’ve searched online for explanatory materials and research papers, but things still aren’t clear to me, so I’m hoping you can help.

Section C.8.8.6.3 describes how inner and outer contours are represented. I don’t have any issues when exporting a torus sliced along its axis: in this case, some slices have one outer and one inner contour, connected by a "bridge," as shown in "torus_along_axis.png."

However, when slicing a torus in a different direction, the resulting slices can have two outer contours, as shown in "torus.png." In this case, do I need to connect those outer contours as well?

Additionally, section C.8.8.6.3 mentions that connections should be oriented around the 12 o’clock position. Is this a mandatory rule, a recommendation, or just an example?

Thanks in advance for your guidance!

[Christof Schadt]

Hi Vitaliy,

no, in that case, these are just two contours, but they must be part of the same ROI.

The 12 o'clock position is just an example where the keyhole is located.

But please also note the second option to not use keyholing, but using an XOR operator. 

In general, most of the systems in Radiotherapy favor the XOR approach creating RT Structure Sets.

But also all of those systems seem to support keyholing when importing the data.

Christof

[Віталій Омельченко]

Hi Christof,

Thank you for your help and advice!

Sincerely,
Vitaliy

середа, 9 липня 2025 р. о 10:03:37 UTC+3 Christof Schadt пише: