Expected input:

• image for each label (file name: use specified name and add label value as suffix; will have no collision in the future as it is guaranteed with T29206 that label values will be unique throughout all groups)
• json that contains the meta info and points to the images (mime type of reader is bound to that meta image)

Way to go for implementation:

• New reader
• New MimeType (Something like "Exported MultiLabelSegmentation")
• Refactor (extract) DICOMSegmentationIO::CreateMetaDataJsonFile so that it can also be used be reused by the new writer
• Refactor (extract) DICOMSegmentationIO::Write Line 159-200 (the code that generates an itk image for each label) so that it can be used by the new writer to generate label images and then directly write them to file.

As soon as the reader is available, we should add a note that mitkFileConvert can than also be used to convert/split DICOM Seg or our MultiLabel nrrd into multiple single label images!!! New Feature: :)