# Grand X-Ray Slam: Division A — Data

来源：https://www.kaggle.com/competitions/grand-xray-slam-division-a/data

采集时间：2026-10-18T13:31:12Z

The Grand X-Ray Slam Division A dataset contains **107,374 chest X-ray images (~138GB)** from three major institutions, designed for **multi-label classification of 14 thoracic conditions**.  

This dataset is **Part 1** of the *ChestDx-MultiInstitution* collection. The other half powers [Grand X-Ray Slam: Division B](https://www.kaggle.com/competitions/grand-xray-slam-division-b).  

Your mission: build AI models for **Dr. HealthAgent** to detect thoracic conditions and assist radiologists worldwide.  

---

##  Files  

- **`train1.csv`** – Training labels for ~107,374 images.  
- **`sample_submission1.csv`** – A sample submission file in the correct format for predictions.  

---

##  How the Data is Organized  
ChestDx-MultiInstitution/
├── train1/
│ ├── 000001_001_001.jpg 
│ ├── 100001_002_001.jpg 
│ ├── 200001_003_001.jpg 
│ └── ...
├── test1/
│ ├── 000301_002_001.jpg
│ └── ...
├── train1.csv
└── sample_submission1.csv


- **train1/** – ~107,374 JPG images (e.g., `000001_001_001.jpg`).  
- **test1/** – ~46,233 JPG test images.  

---

##  Columns in `train1.csv`  

- **Image_Name** – X-ray filename (`PatientID_Study_View.jpg`).  
- **Patient_ID** – Unique anonymized patient identifier
- **Study** – Groups X-rays from the same clinical visit.  
- **Sex** – Male / Female (may contain NaNs).
- **Age** – Patient age (may contain NaNs).  
- **ViewCategory** – Frontal or Lateral.  
- **ViewPosition** – AP / PA / LATERAL.  
- **14 thoracic conditions** (0 = absent, 1 = present):  
  - Atelectasis  
  - Cardiomegaly  
  - Consolidation  
  - Edema  
  - Enlarged Cardiomediastinum  
  - Fracture  
  - Lung Lesion  
  - Lung Opacity  
  - No Finding  
  - Pleural Effusion  
  - Pleural Other  
  - Pneumonia  
  - Pneumothorax  
  - Support Devices  

---

##  What You’re Predicting  

You must detect **14 chest conditions**:  

- **Heart issues** – Cardiomegaly, Enlarged Cardiomediastinum  
- **Lung problems** – Pneumonia, Pneumothorax, Lung Opacity  
- **Fluid-related** – Pleural Effusion, Edema  
- **Other** – Fracture, Support Devices, No Finding  

---

##  Data Provenance

This dataset is a curated merge of multiple public chest X-ray datasets, fully anonymized and preprocessed to ensure no patient overlap between train and test sets.


---

##  Things to Know  

- Images come from different machines and institutions (style variation).  
- Rare conditions (e.g., pneumothorax) → **imbalanced labels**.  
- No patient overlap between train and test.  
- Some **Sex/Age values missing** (`NaN`).  
- Multi-label: one X-ray can show several conditions.  

---

##  Tips  

- Pay attention to **rare conditions** – they’re the toughest.  
- Conditions often **co-occur** (e.g., edema + effusion).  
- Medical context matters – don’t treat each label as fully independent.