YongchengYAO/MSWAL-Lite

About

This is a preprocessed redistribution of MSWAL (HF), which is released under the CC BY-NC 4.0 license.

Dataset summary: 484 abdominal CT scans with 7-class whole-abdominal-lesion masks (gallstone, kidney stone, liver tumor, kidney tumor, pancreatic cancer, liver cyst, kidney cyst).

Contents of this repository:

• Images/ — 484 files
• Masks/ — 484 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	694 single-hospital abdominal CT cases (MICCAI 2025), of which only the 484-case training split was ever uploaded — the 210-case test split listed in the upstream dataset.json points at imagesTs/ files that do not exist on the hub
Excluded here	nothing that exists upstream — all 484 published cases are mirrored
In this repo	484 Images + 484 Masks

All 484 published cases are included. No format conversion was required: the source already ships nii.gz, and image voxel data is carried over unmodified. What is derived here is the flat Images/ + Masks/ layout with aligned basenames (the source's nnU-Net channel suffix _0000 is stripped from image names), the mask re-headering onto the image grid, the uint16 mask cast, and the RAS+ reorientation.

The upstream test split is not withheld here — it was never published: the 210 imagesTs/ entries in the source dataset.json have no corresponding files in the source repository.

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 where needed, 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

• Built from the official HF release zhaodongwu/MSWAL at pinned revision 62c286b05194bfad259de063878355766a6bed9d.

• imagesTr/MSWAL_XXXX_0000.nii.gz is renamed to Images/MSWAL_XXXX.nii.gz so image and mask basenames match; labelsTr/ becomes Masks/.

• The image NIfTI header is copied onto its mask, so each image/mask pair shares one grid and affine.

• Masks are cast to uint16.

• Images and masks are standardized to RAS+ orientation.

📝 The MedVision train/test split (338/146, seed 1024) is a re-split of the 484 published cases; if the upstream authors ever publish their 210-case test split, it will enter MedVision as a new dataset version rather than a rewrite of this one.

Segmentation Labels

labels_map = {
    "1": "gallstone",
    "2": "kidney stone",
    "3": "liver tumor",
    "4": "kidney tumor",
    "5": "pancreatic cancer",
    "6": "liver cyst",
    "7": "kidney cyst"
}

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

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

  • MSWAL_BoxSize_Task01_Axial_Test
  • MSWAL_BoxSize_Task01_Axial_Train
  • MSWAL_BoxSize_Task01_Coronal_Test
  • MSWAL_BoxSize_Task01_Coronal_Train
  • MSWAL_BoxSize_Task01_Sagittal_Test
  • MSWAL_BoxSize_Task01_Sagittal_Train
  • MSWAL_MaskSize_Task01_Axial_Test
  • MSWAL_MaskSize_Task01_Axial_Train
  • MSWAL_MaskSize_Task01_Coronal_Test
  • MSWAL_MaskSize_Task01_Coronal_Train
  • MSWAL_MaskSize_Task01_Sagittal_Test
  • MSWAL_MaskSize_Task01_Sagittal_Train
  • MSWAL_TumorLesionSize_Task01_Axial_Test
  • MSWAL_TumorLesionSize_Task01_Axial_Train
  • MSWAL_TumorLesionSize_Task01_Coronal_Test
  • MSWAL_TumorLesionSize_Task01_Coronal_Train
  • MSWAL_TumorLesionSize_Task01_Sagittal_Test
  • MSWAL_TumorLesionSize_Task01_Sagittal_Train
  • MSWAL_TumorLesionSize_Task02_Axial_Test
  • MSWAL_TumorLesionSize_Task02_Axial_Train
  • MSWAL_TumorLesionSize_Task02_Coronal_Test
  • MSWAL_TumorLesionSize_Task02_Coronal_Train
  • MSWAL_TumorLesionSize_Task02_Sagittal_Test
  • MSWAL_TumorLesionSize_Task02_Sagittal_Train
  • MSWAL_TumorLesionSize_Task03_Axial_Test
  • MSWAL_TumorLesionSize_Task03_Axial_Train
  • MSWAL_TumorLesionSize_Task03_Coronal_Test
  • MSWAL_TumorLesionSize_Task03_Coronal_Train
  • MSWAL_TumorLesionSize_Task03_Sagittal_Test
  • MSWAL_TumorLesionSize_Task03_Sagittal_Train
  • MSWAL_TumorLesionSize_Task04_Axial_Test
  • MSWAL_TumorLesionSize_Task04_Axial_Train
  • MSWAL_TumorLesionSize_Task04_Coronal_Test
  • MSWAL_TumorLesionSize_Task04_Coronal_Train
  • MSWAL_TumorLesionSize_Task04_Sagittal_Test
  • MSWAL_TumorLesionSize_Task04_Sagittal_Train
  • MSWAL_TumorLesionSize_Task05_Axial_Test
  • MSWAL_TumorLesionSize_Task05_Axial_Train
  • MSWAL_TumorLesionSize_Task05_Coronal_Test
  • MSWAL_TumorLesionSize_Task05_Coronal_Train
  • MSWAL_TumorLesionSize_Task05_Sagittal_Test
  • MSWAL_TumorLesionSize_Task05_Sagittal_Train

Data Usage Agreement

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

• You must comply with the original CC BY-NC 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/MSWAL-Lite
• You must cite the original publication(s):
  • https://arxiv.org/abs/2503.13560

Official Release

For more information, please go to the official site: https://github.com/haochen-MBZUAI/MSWAL-

Download from Huggingface

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