README.md

---
configs:
- config_name: default
  data_files:
  - split: train
    path: "modality_dataset.zip"
license: cc-by-4.0
task_categories:
- image-classification
tags:
- medical
- imaging
- ct
- mri
- xray
- ultrasound
pretty_name: Generalized Medical Image Modality Dataset
size_categories:
- 1K<n<10K
---
# 🏥 Generalized Medical Image Modality Dataset

A curated, balanced dataset for training **medical imaging modality classifiers**.
Contains images from four modalities (CT, MRI, X-Ray, Ultrasound) spanning multiple
anatomical regions to ensure robust generalization.

> **Total images:** 9,450 | **Target:** 9,450

---

## 📊 Modality Summary

| Modality | Organ Classes (for Organ Classifier) | Images | Target | % of Total |
|----------|--------------------------------------|-------:|-------:|-----------:|
| CT | Head, Chest, Abdomen | 2,200 | 2,200 | 23.3% |
| MRI | Brain, Spine | 2,000 | 2,000 | 21.2% |
| US | Breast, Kidney, Ovary_Pelvis | 2,250 | 2,250 | 23.8% |
| XRAY | Chest, Hand, Knee | 3,000 | 3,000 | 31.7% |
| **TOTAL** | | **9,450** | **9,450** | **100%** |

---

## 🔬 Per-Organ Breakdown

| Modality | Organ | Images | Target | Kaggle Source | Status |
|----------|-------|-------:|-------:|---------------|--------|
| CT | Abdomen | 1,000 | 1,000 | [nazmul0087/ct-kidney-dataset-normal-cyst-tumor-and-stone](https://www.kaggle.com/datasets/nazmul0087/ct-kidney-dataset-normal-cyst-tumor-and-stone) | ✅ |
| CT | Chest | 1,000 | 1,000 | [mohamedhanyyy/chest-ctscan-images](https://www.kaggle.com/datasets/mohamedhanyyy/chest-ctscan-images) | ✅ |
| CT | Head | 200 | 200 | [felipekitamura/head-ct-hemorrhage](https://www.kaggle.com/datasets/felipekitamura/head-ct-hemorrhage) | ✅ |
| MRI | Brain | 1,000 | 1,000 | [masoudnickparvar/brain-tumor-mri-dataset](https://www.kaggle.com/datasets/masoudnickparvar/brain-tumor-mri-dataset) | ✅ |
| MRI | Spine | 1,000 | 1,000 | [anoukstein/spider-mri-spine-t2-png](https://www.kaggle.com/datasets/anoukstein/spider-mri-spine-t2-png) | ✅ |
| US | Breast | 750 | 750 | [aryashah2k/breast-ultrasound-images-dataset](https://www.kaggle.com/datasets/aryashah2k/breast-ultrasound-images-dataset) | ✅ |
| US | Kidney | 750 | 750 | [gurjeetkaurmangat/kidney-ultrasound-images-stone-and-no-stone](https://www.kaggle.com/datasets/gurjeetkaurmangat/kidney-ultrasound-images-stone-and-no-stone) | ✅ |
| US | Ovary_Pelvis | 750 | 750 | [orvile/mmotu-ovarian-ultrasound-images-dataset](https://www.kaggle.com/datasets/orvile/mmotu-ovarian-ultrasound-images-dataset) | ✅ |
| XRAY | Chest | 1,000 | 1,000 | [paultimothymooney/chest-xray-pneumonia](https://www.kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia) | ✅ |
| XRAY | Hand | 1,000 | 1,000 | [antonbudnychuk/hand-xray](https://www.kaggle.com/datasets/antonbudnychuk/hand-xray) | ✅ |
| XRAY | Knee | 1,000 | 1,000 | [shashwatwork/knee-osteoarthritis-dataset-with-severity](https://www.kaggle.com/datasets/shashwatwork/knee-osteoarthritis-dataset-with-severity) | ✅ |

---

## 📁 Folder Structure

```
modality_dataset/
├── CT/
│   ├── Chest/        (CT chest scans)
│   ├── Head/         (CT head hemorrhage scans)
│   └── Abdomen/      (CT kidney / abdominal scans)
├── MRI/
│   ├── Brain/        (Brain tumor MRI)
│   └── Spine/        (Spine T2 MRI)
├── XRAY/
│   ├── Chest_Pneumonia/  (Chest X-ray — pneumonia)
│   ├── Chest_TB/         (Chest X-ray — tuberculosis)
│   └── Hand/             (Hand X-ray)
└── US/
    ├── Breast/       (Breast ultrasound)
    ├── Kidney/       (Kidney ultrasound)
    └── Ovary_Pelvis/ (Ovarian / pelvic ultrasound)
```

## 🚀 Usage

Images are packaged in `modality_dataset.zip`. Extract and use directly:

```python
import zipfile
with zipfile.ZipFile("modality_dataset.zip", "r") as z:
    z.extractall("./modality_dataset")
```

The top-level folder names (`CT`, `MRI`, `XRAY`, `US`) are the **class labels** for
the modality classifier. Sub-folders represent the anatomical regions used as data sources.