Markus D. Herrmann

> If the second approach is correct I should add some code to throw an error if it encounters input from the first approach. I think I have implemented the...

@seandoyle we could now implement subclasses of `ImageLibraryEntryDescriptors` for specific imaging modalities (TID 160{3-7}).

I am not sure whether we should implement TID 1602 Image Library Entry or whether we should omit this for now and only support descriptions of groups of images. The...

Ok. Thanks @fedorov and @seandoyle. If I understand it correctly, the group-level descriptors are intended to provide attributes that are common to the collection of image instances and TID 1601...

> This would certainly simplify the code that is using highdicom. To avoid a lot of conditional logic (and standardize the SR creation) should we assume that all of the...

This is a tricky question. Generally, I would argue that the library has to assume passed data sets are standard compliant and I would be cautious about just "fixing" things...

Yes, @CPBridge is right. I wanted to strongly discourage the use of free text for qualitative evaluations.

Another alternative would be use a private (potentially project-specific) coding system. I would prefer that over using TEXT. What do you think @fedorov @pieper @CPBridge @seandoyle?

> Using codes (private or public) assumes that you know what you expect to find in advance. Using private codes just eliminates the burden of finding a standard code. How...

> Implementing this in a backwards compatible way could get a bit messy. We'd probably need to create a new, alternative `TextQualitativeEvaluation`, or similar, and then accept `Sequence[Union[QualitativeEvaluation, TextQualitativeEvaluation]]` for...