Hugging Face's logo

Datasets:
FangDai
/
Multimodal_Lung_Cancer_Dataset-CT-Mask-Clinical

License:
Dataset card Files
xet

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

You agree to not use the dataset to conduct experiments that cause harm to human subjects.

Log in or Sign Up to review the conditions and access this dataset content.

Lung Cancer Multimodal Imaging Dataset for Treatment Response and Survival Analysis

📌 Overview

This dataset provides a curated multimodal imaging and clinical cohort of lung cancer patients undergoing chemotherapy or immunotherapy, designed to support research in:

  • Lung CT-based segmentation and representation learning
  • PET/CT multimodal modeling
  • Radiomics-driven outcome prediction
  • Survival analysis (OS / DFS)
  • Treatment response stratification
  • Histological subtype classification (LUAD vs LUSC)

The dataset is organized at the patient level, with aligned imaging, clinical, and outcome data to facilitate reproducible AI and statistical modeling studies.

📁 Dataset Structure 

PublicData/
├── chemo-CT/
│   ├── chemoct1_001_ct_image/ (image & mask)
│   ├── chemoct1_002_ct_image/ (image & mask)
│   ├── chemoct1_003_ct_image/ (image & mask)
│   ├── chemoct1_004_ct_image/ (image & mask)
│   └── chemoct1_005_ct_image/ (image & mask)
│
├── chemo-PET/
│   ├── chemopet1_001_pet_image.nrrd
│   ├── chemopet1_002_pet_image.nrrd
│   ├── chemopet1_003_pet_image.nrrd
│   ├── chemopet1_004_pet_image.nrrd
│   └── chemopet1_005_pet_image.nrrd
│
├── immune-CT/
│   ├── immunect_004/ (image & mask)
│   ├── immunect_005/ (image & mask)
│   ├── immunect_006/ (image & mask)
│   ├── immunect_009/ (image & mask)
│   └── immunect_010/ (image & mask)
│
├── immune-PET/
│   ├── immunepet_004_pet_image.nrrd
│   ├── immunepet_005_pet_image.nrrd
│   ├── immunepet_006_pet_image.nrrd
│   ├── immunepet_009_pet_image.nrrd
│   └── immunepet_010_pet_image.nrrd
│
└── clinical.xlsx

Imaging Data Description

CT Imaging

  • Format: PNG (2D slices)

  • Modality: Chest CT

  • Organization: Slices grouped per patient

  • Intensity: Linearly normalized from original HU values (no windowing applied)

  • Resolution: Consistent within each patient

PET Imaging

  • Format: NRRD (3D volumes)

  • Modality: FDG-PET

  • Stored as one volume per patient

  • Spatially aligned to CT at the patient level (not voxel-level registered)

🫁 Lung Mask & Radiomics Feature Generation

  • Lung segmentation masks were generated using Model TotalSegmentator, a previously published and publicly available lung segmentation model.
    No manual correction was applied, and the masks are provided as-is for pipeline testing and reproducibility purposes.
  Wasserthal, J., Breit, H.-C., Meyer, M.T., Pradella, M., Hinck, D., Sauter, A.W., Heye, T., Boll, D., Cyriac, J., Yang, S., Bach, M., Segeroth, M., 2023. TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images. Radiology: Artificial Intelligence. https://doi.org/10.1148/ryai.230024

  • Radiomics features were extracted using the following Python packages:

    • PyRadiomics (https://github.com/AIM-Harvard/pyradiomics)
      – for standardized handcrafted radiomic feature extraction (first-order, shape, texture features)
    • SimpleITK
      – for image I/O, resampling, and preprocessing

  • Feature extraction followed default PyRadiomics settings unless otherwise specified, ensuring full reproducibility across environments.

⚠️ The provided lung masks and radiomics features are intended only for methodological validation and code testing, not for clinical use or performance benchmarking.

📊 Clinical and Outcome Data (clinical.xlsx)

The clinical spreadsheet contains one row per patient, including: 

- Age
- Sex
- Smoking status (Yes / No)
- Smoking duration
- Smoking cessation status
- Tumor diameter
- Treatment Information
- Treatment group: Chemotherapy or Immunotherapy
- Survival Outcomes
- Overall Survival (OS): time (months), event (0/1)
- Disease-Free Survival (DFS): time (months), event (0/1)
- Labels
- Histological subtype:
- 0 = Lung Adenocarcinoma (LUAD)
- 1 = Lung Squamous Cell Carcinoma (LUSC)

🔗 Patient ID Convention

All data modalities are linked via a consistent patient identifier: 

Modality	Example
CT folder	immunect_004/
PET file	immunepet_004_pet_image.nrrd
Clinical record	Patient_ID = 004

This enables direct matching across:

  • CT slices

  • PET volumes

  • Clinical and survival data

🧪 Intended Use Cases

This dataset is released as a minimal test set for GitHub Model A, designed to facilitate quick verification, reproducibility checks, and baseline benchmarking. It is not intended for full model training, but rather for validating core functionalities and end-to-end pipelines.

https://github.com/fangdai-dear/MM-DLS.git

Specifically, this dataset can be used to:

  • Validate lung and lesion segmentation pipelines

  • Perform sanity checks for multimodal CT (±PET) representation learning

  • Test radiomics feature extraction and integration workflows

  • Verify AI-based survival prediction pipelines (input–output consistency)

  • Conduct treatment response stratification demos

  • Benchmark LUAD vs LUSC classification in a controlled, lightweight setting

  • Serve as a minimal benchmark for deep learning models (e.g., UNet, Attention UNet, nnU-Net, MONAI) to ensure correct implementation and reproducibility

🔄 Reproducibility & Simulation

For reproducibility and benchmarking:

  • Imaging data are provided in standardized formats

  • Clinical variables are structured and anonymized

  • The dataset supports simulated data augmentation and survival modeling

  • Example utilities for simulation and evaluation can be built on top of this structure

📜 License & Data Usage

This dataset is released for non-commercial research use.
If you use this dataset in academic work, please cite the associated publication or repository as specified on the hosting platform (GitHub). 
https://github.com/fangdai-dear/MM-DLS.git
Data Usage Restriction: The associated manuscript is currently under review and has not yet been published. Use of this dataset for any purpose beyond code testing, reproducibility checks, or private evaluation is strictly prohibited prior to the official publication of the article.

📬 Contact

For questions, issues, or collaboration inquiries, please contact:

  • Dataset Maintainer

  • Email: [lxy1753565@163.com]

  • Affiliation: [Shanghai Pulmonary Hospital, Tongji University Medical ]

  • Version: v1.0

  • Last Updated: 2025

Downloads last month
9