dataset.get() cannot read tags with 4 elements about dicom-anonymizer

[laurennlam]

I need to go deeper on Dicom tags to understand more this case.
Thanks!

[laurennlam]

After some reasearches, this kind of tag has been introduced in the library in order to manage Repeating Groups.
Indeed, this case is not handled in the anonymize_dataset function for private tags. I'll make a pull request for it. Also, it's a real issue if we change the default associated action to these tags. Today, there are deleted. The delete function manages the case with 4 tags but it's the only one.
That's why the anonymize_dataset function needs to be updated in order to not adding this case in all actions methods.

So, to conclude, having 4 values in a tag is not a dicom convention but specific to this library.
Thanks for pointed it out.

[laurennlam]

Thanks for sharing this issue.
Do you have a file you can share so that I can check this on my side ?

[LudvigOlsen]

Hi Laurenn,

I don't have a file, I can share no. Does this not happen with the data you are testing with?
I think pydicom might have a way to make a dummy image to test with.

[laurennlam]

I don't have any file with this kind of tag.
It seems that sometime, we handle this case :

def delete(dataset, tag):
    """X - remove"""

    def range_callback(dataset, data_element):
        if data_element.tag.group & tag[2] == tag[0] and data_element.tag.element & tag[3] == tag[1]:
            delete_element(dataset, data_element)

    if len(tag) > 2:  # Tag ranges
        dataset.walk(range_callback)

When do you have the issue ? Can you describe more the context ?

[LudvigOlsen]

Right, my files did not have them either. But because these tags are part of the dicomfields.py lists, get() is called on it and fails. So I get the error message for all files (I call on one file at a time btw.). I believe the solution in my PR would work (although perhaps a bit hacky).
I should say that I'm no longer working on this project.