LIDC-annot-NLST501 — Measurement Units and CT/Segmentation Geometry Clarification

Support
1

Hello IDC team,

I am a Master’s researcher using the LIDC-annot-NLST501 dataset for pulmonary nodule size validation.

After reviewing the official TCIA documentation and examining the DICOM SEG metadata, I would appreciate clarification on three points:

  1. Measurement units: The documentation defines extracted_lax, extracted_sax, and extracted_lad in mm, and extracted_volume in mm³. However, our reconstructed mask measurements show closer numerical agreement with pixel lengths and voxel counts. Has this discrepancy been reported or investigated?

  2. Slice alignment: Is there an officially recommended method for mapping location_z / extracted_z from annotations.csv to the original NIfTI masks and DICOM CT slices? Should the indexing follow NIfTI array order, or be determined through physical image geometry?

  3. Corrected CT series: We understand that 24 CT series were corrected. Are there any known issues involving slice counts, SEG references, or spatial alignment in this release?

If available, could you direct us to reference conversion code, known issues, or documentation addressing these questions?

Thank you for your support.

2

Nadine, thank you for reaching out about this issue!

This specific collection is provided to IDC by The Cancer Imaging Archive (TCIA) (source location: LIDC-annot-NLST501 - The Cancer Imaging Archive (TCIA)), one of the IDC data sources. All questions about the content of that collection should be raised with the TCIA helpdesk.

For your convenience, I have just opened the issue with them, and asked them to add you by your email that you used to register on this forum. You may need to create an account on TCIA help desk website first. This is the issue I created: https://help.cancerimagingarchive.net/servicedesk/customer/portal/1/TH-60532.

I am also tagging Justin Kirby @kirbyju, one of the TCIA leads, for awareness.