DICOM: RTSTRUCT not within the proper study in models hierarchy

[Elaine Meng]

For Head-Neck Cetuximab collection, patient ID 0522c0027, there are 3 subdirectories corresponding to 3 different studies in Jan, Feb, and May (subdir name is derived from date). The RTSTRUCT are within the Feb subdirectory but ChimeraX opens them as subdirectories of a "date unknown" model, rather than of the 20000220 (Feb 20, 2000) model, which would be the correct thing to do. The directory hierarchy should be maintained in the model hierarchy so that it is clear which CT goes with these RTSTRUCT models.

I tried to make a zip file of just this patient but I recall there may be some problems with using the Mac "compress directory" function (which is what I used). So if the attached zip file doesn't work you may need to use the TCIA website and its download app instead. The page for this study is ​https://wiki.cancerimagingarchive.net/display/Public/Head-Neck+Cetuximab

I didn't see a DICOM component so I made it Input/Output. Modify this ticket as desired.

[goddard@…]

Good to note this problem.  But I suggest not working on it for now because time on DICOM should be focussed on figuring out Jason Rubenstein's needs which probably do not involve this RTSTRUCT (radio-therapy) type of data.  In the future if we have interest in RT data that would be the time to fix this.

[Elaine Meng]

I could not attach the zip file because it was too big. However, example data on plato /wynton/group/ferrin/usr.local/projects/chimerax/www/data/dicom/02-20-2000-522-26564.zip (when unzipped) contains just the one study from February that includes the RTSTRUCT models.  So if you open that you can see that in ChimeraX, the 20000220 model (Feb study)  is #1.1.1 but the RTSTRUCT stuff goes under "date unknown" #1.1.2.  

[Elaine Meng]

Note that the rtstruct data is actually already inside the directory with the proper study date, so fixing this simply means to keep the hierarchy that is already present in the input directories.  It does not require reading the files to glean some information from metadata.

[goddard@…]

If I recall correctly, the directory names and organization are not used and are not part of the DICOM standard.  The grouping comes from the metadata in the files, not from what directory files were found in.  The "unknown date" for RTSTRUCT probably reflects some missing metadata.

[Elaine Meng]

If so, it did not stop Horos and Slicer from associating it automatically with the correct study date.

[Zach Pearson]

We can infer the study date from the series date, as Slicer and Horos apparently do (it's the only way they could possibly make this association).

[Zach Pearson]

This will be fixed when I merge my DICOM metadata browser.