Spot Check for PET Required Tags in LTP2

[hoodriverheather]

@nutjob4life
ChatGPT is saying that these PET required tags might exist in the header, but maybe either nested within the Radiopharmaceutical Information Sequence (0054,0016). Or they might exist at the Series level, but not in each PET DICOM file.

Would you please check the PET DICOM files for this site/event?

SiteID: ldytNSGnHnrBQ
EventID: 1165313

We are seeing repeated missing PET findings for:

Radionuclide Total Dose (0018,1074)
Radionuclide Half Life (0018,1075)
Radiopharmaceutical Start Time (0018,1072)

Could you inspect 1 to 3 sample files from the same affected PET series, for example a first, middle, and later image, and check whether:

Radiopharmaceutical Information Sequence (0054,0016) is present
those PET fields are present within that sequence item

Sample file path to start with:
DICOM/I456

If those values are present inside Radiopharmaceutical Information Sequence, then the validator likely needs to traverse the nested PET metadata correctly.

If those values are absent there as well across multiple files from the same series, then this is likely a legacy/export metadata limitation rather than a validation parsing issue.

[nutjob4life]

@hoodriverheather since my to-do list is absolutely dense today and the need for this answer is "by today", I had AI do the majority of the work. I haven't checked its work. Here is its research and conclusion:

.venv/bin/spot-check-pet-tags ../original-data/Lung_Team_Project_2/Images_Site_ldytNSGnHnrBQ/1165313
Series 1.3.6.1.4.1.5962.99.1.3544515373.1186743414.1... (213 instances, Modality=PT)
  Sample I456 (InstanceNumber=1):
    (0054,0016) RadiopharmaceuticalInformationSequence: root=present (1 item)
    (0018,1074) RadionuclideTotalDose: root=missing | nested=370000000
    (0018,1075) RadionuclideHalfLife: root=missing | nested=6586.199707
    (0018,1072) RadiopharmaceuticalStartTime: root=missing | nested=103400
    (0018,1078) RadiopharmaceuticalStartDateTime: root=missing | nested=20140725103409
  Sample I562 (InstanceNumber=107):
    (0054,0016) RadiopharmaceuticalInformationSequence: root=present (1 item)
    (0018,1074) RadionuclideTotalDose: root=missing | nested=370000000
    (0018,1075) RadionuclideHalfLife: root=missing | nested=6586.199707
    (0018,1072) RadiopharmaceuticalStartTime: root=missing | nested=103400
    (0018,1078) RadiopharmaceuticalStartDateTime: root=missing | nested=20140725103409
  Sample I668 (InstanceNumber=213):
    (0054,0016) RadiopharmaceuticalInformationSequence: root=present (1 item)
    (0018,1074) RadionuclideTotalDose: root=missing | nested=370000000
    (0018,1075) RadionuclideHalfLife: root=missing | nested=6586.199707
    (0018,1072) RadiopharmaceuticalStartTime: root=missing | nested=103400
    (0018,1078) RadiopharmaceuticalStartDateTime: root=missing | nested=20140725103409
  Conclusion: LIKELY_VALIDATOR_PARSING_ISSUE
  Required PET tags exist inside RadiopharmaceuticalInformationSequence on all inspected samples
  but are absent at the dataset root; the validator likely needs to traverse nested PET metadata.

[hoodriverheather]

@nutjob4life It looks like there are values in the nested PET metadata and the validator will need to "traverse" to read it. I think I need to update the EDRN DICOM tags to include that logic for you - correct?

[nutjob4life]

@hoodriverheather yes that sounds correct.

[hoodriverheather]

@nutjob4life I updated the PET metadata tags to include specifications for "traversing" nested tags. Please review in this updated copy of the EDRN Core DICOM Tags sheet.

this version should also replace the DRAFT version that is linked to the Portal. What's a good process for this?

Let me know if you want a quick call to discuss.
Thanks!

[nutjob4life]

@hoodriverheather wrote:

this version should also replace the DRAFT version that is linked to the Portal. What's a good process for this?

Let's talk about this during the staff call today, 2026-07-14, at 2pm PT.

[nutjob4life]

@hoodriverheather, I mentioned above

Let's talk about this during the staff call today, 2026-07-14, at 2pm PT.

But we ran out of time to discuss this! I blame Dan 😉

I manually updated the "official" copy from your DRAFT version.

In the future:

• We could automate this by having the "official" public copy automatically import the DRAFT; whenever there's a change to the draft, it'll appear in the offical
  • But at this point, we may as well drop the DRAFT and just do all changes on the "official" public copy in the first place
• I could share the edrn-bot credentials so you can update the "official" public copy yourself
• You could use the "Share ▶ Publish to Web" feature to turn the spreadsheet into web pages for the portal to use

Let me know what you prefer.

[hoodriverheather]

@nutjob4life Haha! Thanks for updating the portal.
I would like to keep two sheets. I just updated the DRAFT to EDRN LabCASMetadata05242024-Working Copy v1.1 Draft so that I can keep track of proposed changes and then push them to the "official" public copy when everyone agrees.
I hope we can move to LinkML soon and keep everything in GitHub.

[nutjob4life]

@hoodriverheather I have updated the validator program (or "script" if you like) with the nested Radiopharmaceutical Information Sequence tag handling. I would consider it a first draft as I'm not sure if it works. Let me know if you'd like to test it or what the next steps are.

[hoodriverheather]

@nutjob4life I can test if you would provide output for LTP2. Would it be easier if I gave you a specific site to run?

[nutjob4life]

@hoodriverheather LTP2 (Lung_Team_Project_2) is fairly large (571,436 files at 296.2 GB) so I will start the scan immediately. However it may take a full day (or more) for the results to be ready. I'll ping you here on this issue when it's done.

Would it be easier if I gave you a specific site to run?

It would definitely save time. Let me know what your preferred site is. In the meantime the full Lung_Team_Project_2 will run overnight.