SegmentedPropertyTypeModifier seems to be ignored while loading terminology from JSON

[fedorov]

SegmentedPropertyTypeModifier is not showing up in the GUI, although it looks like it is present in the JSON file produced by the dcmqi converter.

Test dataset:

• MR series: http://slicer.kitware.com/midas3/folder/4434
• SEG series: http://slicer.kitware.com/midas3/folder/4435 (any of the files in this folder will do)

Output of segimage2itkimage (relevant part):

            "SegmentedPropertyTypeCodeSequence" : {
               "CodeMeaning" : "Cerebral hemisphere",
               "CodeValue" : "T-A010F",
               "CodingSchemeDesignator" : "SRT"
            },
            "SegmentedPropertyTypeModifierCodeSequence" : {
               "CodeMeaning" : "Right",
               "CodeValue" : "G-A100",
               "CodingSchemeDesignator" : "SRT"
            },

Presentation in Slicer Segmentations module (modifier not initialized):

[cpinter]

Is it maybe an update issue? What happens if you click another property then back?

[cpinter]

I'm trying to build DCMQI so that I can try it (I have a build error, but I'll try some things first).

In the meantime a question: you mention in the title that the terminology was loaded from Json, but I see no Json file. Can you attach that? Or you mean the segment descriptor Json that the converter creates from the SEG files?

[fedorov]

Or you mean the segment descriptor Json that the converter creates from the SEG files?

Yes, I just looked at the JSON file that was produced by the converter.

If you use the current master of Slicer, you will need to make sure you use the latest master of dcmqi, since otherwise there will be an API mismatch with dcmtk, due to the upgrade of dcmtk in Slicer last night.

[cpinter]

Yes, I updated DCMQI but then I realized I need to rebuild DCMTK

[cpinter]

I updated and built eberything, and loaded the dataset. I confirm that there is no type modifier showing up. However I canont find the json file. There is a folder created c:\Users\pinter\AppData\Local\Temp\QIICR\SEG\2017-04-28_114148 and I think it should be there, because that's where the json files were put before, and the date in the folder name indicates it should contain it. But it's all empty. Any ideas?

[fedorov]

I think I experienced the same issue. I don't know if the temp dir is removed by our code, or something else, and didn't want to spend time investigating, so I just took the command line from the log, and re-ran it manually to get the JSON.

Now that I looked into the code, it does appear that the temp dir is cleaned up upon completion: https://github.com/QIICR/QuantitativeReporting/blob/master/QuantitativeReporting.py#L549-L553.

[cpinter]

Thanks! Probably I'll just comment that out for debugging.

[cpinter]

I could work on this a bit. The cleanupTemporaryData function is called when saving a report, not when loading a SEG. @che85 If you have any comments, please don't hold it back. You can easily save half an hour for me by a half-minunte comment :) Thanks

[fedorov]

@cpinter Christian is on vacation until the end of this week. If you like, we can wait until he returns? Next week is fine, there is no extreme urgency.

[cpinter]

I can keep trying for a bit more. I managed to quickly save the json file, but it did not contain modifiers. Can you tell me which series is the one you had the json for? Thanks!

[cpinter]

Found it, nevermind! left_right_gm_wm_snomedhas modifiers.

[fedorov]

right - sorry - only those that have "snomed" in the name contain modifiers!

[cpinter]

One more question (sorry for the flood!): Looking at the logic code, modifiers were supported by looking for Modifier tags under SegmentedPropertyTypeCodeSequence. Maybe the convention has been changed sometime in the past, but the info was not propagated. Or just another faulty assumption on my part... Should we support more types of mata.json structures, or should I just make it handle the newest one?

[fedorov]

No worries re flood :)

I don't think this part changed, but of course I could be wrong. Modifiers are supposed to be at the same level as the category/type. The organization is defined by this section of the schema: https://github.com/QIICR/dcmqi/blob/master/doc/schemas/seg-schema.json#L109-L141. I think fixing modifier is sufficient for now.

[cpinter]

You're right, after another look, the code also suggests it was always like that. So my job will be somewhat harder than it looked a few mintes ago :)

[cpinter]

It was a simple fix at the end: https://github.com/Slicer/Slicer/commit/1c4e327436dc280dad4fddff8a2f47a714fdc2a3

Type modifier is now correctly populated, at least on my end. Please confirm on yours.

[fedorov]

Nice! I will check in the tomorrow's nightly, since I don't have a recent Slicer build. Thank you!

[fedorov]

Confirmed the issue is resolved in today's nightly. Thank you!