DICOM RT structure set can only store simple contours, not such complex segmentations as a FreeSurfer brain segmentation.
DICOM Segmentation Object can store arbitrarily complex segmentation. However, it requires you to refer to a reference image and that reference image must have UIDs for each frame.
If you don’t have the input DICOM image anymore then you can create a dummy image by the following steps:
- export the segmentation to a labelmap volume
- convert the labelmap volume to scalar volume (using Volumes module → Volume Information → Convert to scalar volume)
- export the volume to DICOM using scalar volume exporter (UIDs are generated during export), and loading that exported image
- load the image from the DICOM database
We could certainly make things simpler by automatically generating dummy volumes, but since segmentation has limited use without a real referenced volume, there has not been many requests for this.
Nifti file format cannot store segment names in the file, therefore you need to use an external colormap file. If you use freesurfer then you can load the segmentation as a volume, then in Volumes module choose the freesurfer colormap, and then convert it to segmentation.