YongchengYAO/LNQ2023-Lite

About

This is a preprocessed redistribution of LNQ2023 / TCIA collection MEDIASTINAL-LYMPH-NODE-SEG (DOI 10.7937/QVAZ-JA09), which is released under the CC BY 4.0 license.

Dataset summary: 120 chest CT scans with EXHAUSTIVELY annotated mediastinal lymph node segmentation masks.

Contents of this repository:

• Images/ — 120 files
• Masks/ — 120 files

📝 Landmark annotations, visualization figures and the benchmark plan files live in 🔥MedVision🔥, where you can load the complete images and annotations from dataset configs.

Relation to the source dataset

In the source	513 chest CT series with DICOM-SEG annotations — 120 fully annotated, 393 partially annotated
Excluded here	the 393 partially annotated cases (most of their lymph nodes are unlabelled)
In this repo	120 Images + 120 Masks

Only exhaustively annotated cases are kept: 513 → 120.

Each DICOM-SEG series in the TCIA release declares its own completeness in the DICOM SeriesDescription tag — either Fully Annotated (120 series) or Partially Annotated (393 series). The partially annotated cases are the challenge's training set, where only a subset of the visible lymph nodes was contoured.

That gap is large, and it is visible in the masks themselves:

	cases	nodes/case (mean)	median	max	cases with exactly 1 node
Fully Annotated	120	9.00	8	42	1 / 120 (1%)
Partially Annotated	393	1.46	1	6	249 / 393 (63%)

A 6.2× difference: 63% of partially annotated cases carry exactly one segmented node, against a median of 8 for fully annotated ones — so most true nodes in those cases are simply unlabelled. Unlabelled is not negative: a model that correctly detects such a node would be scored as a false positive, and a size measurement on it has no reference value. Those cases cannot serve as benchmark ground truth, so they are not redistributed.

Why -Lite? The suffix marks this as a derived redistribution rather than a copy of the source. These are preprocessed volumes — every case has been format-converted, geometry-normalised and reoriented to RAS+ — and for some sources cases or modalities are excluded as well (see the table above). Use it to reproduce MedVision, not as a substitute for the original release. See Preprocessing below for exactly what was changed.

Preprocessing

• DICOM CT series converted to nii.gz; DICOM-SEG objects decoded to label masks.

• CT and SEG are paired via ReferencedSeriesSequence rather than by filename order.

• Images and masks standardized to RAS+ orientation.

• Only SEG series whose DICOM SeriesDescription is Fully Annotated are converted; the 393 partially annotated series are skipped at download time.

📝 Redistributed from the TCIA release (CC BY 4.0), not the Zenodo challenge copy (CC BY-NC-ND).

Segmentation Labels

labels_map = {
    "1": "mediastinal lymph node"
}

Landmarks

landmarks_map = {
    "P1": "most right/anterior/superior endpoint of the major axis",
    "P2": "most left/superior/inferior endpoint of the major axis",
    "P3": "most right/anterior/superior endpoint of the minor axis",
    "P4": "most left/superior/inferior endpoint of the minor axis"
}

News

• [25 Jul, 2026] Initial release. This dataset is integrated into 🔥MedVision🔥, where you can use these config names to load data in python:

  • LNQ2023_BoxSize_Task01_Axial_Test
  • LNQ2023_BoxSize_Task01_Axial_Train
  • LNQ2023_BoxSize_Task01_Coronal_Test
  • LNQ2023_BoxSize_Task01_Coronal_Train
  • LNQ2023_BoxSize_Task01_Sagittal_Test
  • LNQ2023_BoxSize_Task01_Sagittal_Train
  • LNQ2023_MaskSize_Task01_Axial_Test
  • LNQ2023_MaskSize_Task01_Axial_Train
  • LNQ2023_MaskSize_Task01_Coronal_Test
  • LNQ2023_MaskSize_Task01_Coronal_Train
  • LNQ2023_MaskSize_Task01_Sagittal_Test
  • LNQ2023_MaskSize_Task01_Sagittal_Train
  • LNQ2023_TumorLesionSize_Task01_Axial_Test
  • LNQ2023_TumorLesionSize_Task01_Axial_Train

Data Usage Agreement

By using the dataset, you agree to the terms as follow.

• You must comply with the original CC BY 4.0 license terms of the source dataset.
• You are recommended to refer to the source of this dataset in any publication: https://huggingface.co/datasets/YongchengYAO/LNQ2023-Lite
• You must cite the original publication(s):
  • https://doi.org/10.7937/QVAZ-JA09

Official Release

For more information, please go to the official site: https://lnq2023.grand-challenge.org/

Download from Huggingface

# python
from huggingface_hub import snapshot_download
snapshot_download(repo_id="YongchengYAO/LNQ2023-Lite", repo_type='dataset', local_dir="/your/local/folder")