Angelou0516/ARCADE

---
dataset_info:
- config_name: syntax
features:
- name: image
dtype: image
- name: mask
dtype: image
- name: image_id
dtype: int64
- name: file_name
dtype: string
splits:
- name: train
num_examples: 1000
- name: val
num_examples: 200
- name: test
num_examples: 300
- config_name: stenosis
features:
- name: image
dtype: image
- name: mask
dtype: image
- name: image_id
dtype: int64
- name: file_name
dtype: string
splits:
- name: train
num_examples: 1000
- name: val
num_examples: 200
- name: test
num_examples: 300
configs:
- config_name: syntax
data_files:
- split: train
path: syntax/train-*
- split: val
path: syntax/val-*
- split: test
path: syntax/test-*
- config_name: stenosis
data_files:
- split: train
path: stenosis/train-*
- split: val
path: stenosis/val-*
- split: test
path: stenosis/test-*
license: cc0-1.0
task_categories:
- image-segmentation
tags:
- medical-imaging
- coronary-artery
- x-ray-angiography
- SYNTAX-score
pretty_name: ARCADE
size_categories:
- 1K<n<10K
---

# ARCADE

Automatic Region-based Coronary Artery Disease diagnostics using X-ray angiography imagEs.

## Description

X-ray coronary angiography (XCA) dataset for coronary artery segmentation and stenosis detection. Contains 3,000 grayscale 512x512 images across two tasks.

## Subsets

- **syntax**: Coronary artery segment classification into 26 anatomical regions per SYNTAX Score methodology (1,500 images)
- **stenosis**: Atherosclerotic plaque / stenotic lesion detection and segmentation (1,500 images)

## Splits

| Split | Images per subset |
|-------|-------------------|
| Train | 1,000 |
| Val | 200 |
| Test | 300 |

## Mask Labels

**Syntax** (26 classes): Coronary artery segments 1-16c per SYNTAX Score methodology.

**Stenosis** (26 classes): Same artery segments (1-25) plus class 26 = stenosis lesion.

Pixel value 0 = background.

## Source

- **Paper**: Popov et al., "Dataset for Automatic Region-based Coronary Artery Disease Diagnostics Using X-Ray Angiography Images", *Scientific Data* 11, 20 (2024). DOI: 10.1038/s41597-023-02871-z
- **Zenodo**: https://zenodo.org/records/10390295
- **License**: CC0 1.0 Universal