Angelou0516/WSSS4LUAD-v2

WSSS4LUAD Dataset

This is the WSSS4LUAD dataset for weakly-supervised tissue semantic segmentation in lung adenocarcinoma.

Dataset Description

The dataset contains histopathology image patches from lung adenocarcinoma tissue samples, designed for weakly-supervised semantic segmentation tasks.

Dataset Structure

• Training: 10,091 patches with image-level labels
• Validation: 40 patches with pixel-level masks
• Test: 80 patches with pixel-level masks

Classes

The dataset includes 3 tissue classes:

1. Tumor epithelial tissue (label_tumor)
2. Tumor-associated stroma tissue (label_stroma)
3. Normal tissue (label_normal)

Features

• image_id: Unique identifier for each image
• image: The histopathology image patch (PNG format, embedded)
• filename: Original filename
• width: Image width in pixels
• height: Image height in pixels
• split: Dataset split (train/validation/test)
• label_tumor: Binary label for tumor epithelial tissue (training only)
• label_stroma: Binary label for tumor stroma tissue (training only)
• label_normal: Binary label for normal tissue (training only)
• mask: Pixel-level segmentation mask (validation/test only)
• background_mask: Background region mask (validation/test only)

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("USERNAME/WSSS4LUAD")

# Access different splits
train_data = dataset['train']
val_data = dataset['validation']
test_data = dataset['test']

# Example: Load and display an image
example = train_data[0]
image = example['image']  # PIL Image
labels = {
    'tumor': example['label_tumor'],
    'stroma': example['label_stroma'],
    'normal': example['label_normal']
}

Citation

If you use this dataset, please cite the original WSSS4LUAD paper.

License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.