---
license: odc-by
pretty_name: LoDoPaB-CT (GLIMPSE subsets)
tags:
- computed-tomography
- medical-imaging
- inverse-problems
task_categories:
- image-to-image
---

# LoDoPaB-CT subsets for GLIMPSE

Processed image subsets used by
[GLIMPSE](https://github.com/swing-research/Glimpse)
([paper](https://arxiv.org/abs/2401.00816)). Sinograms are **not** stored; they
are rendered on the fly by the GLIMPSE data pipeline (ODL / scikit-image).

## Layout

| Split | Contents | Format |
|---|---|---|
| `train/` | LoDoPaB-CT training slices | `.npy` float32 arrays |
| `test/`  | LoDoPaB-CT test slices | `.npy` float32 arrays |
| `ood/`   | out-of-distribution brain images | `.jpg` |

```python
import numpy as np  # a .npy slice
img = np.load("train/0.npy")        # (H, W) float32 in [0, 1]
```

## License & attribution

This repo mixes two sources with different (but both attribution-only) licenses;
credit both when reusing.

**`train/` + `test/` — LoDoPaB-CT** (Leuschner et al., *Scientific Data* 2021),
**ODC-By v1.0**, DOI [10.5281/zenodo.3384092](https://doi.org/10.5281/zenodo.3384092).
Built on **LIDC-IDRI** from [TCIA](https://www.cancerimagingarchive.net/collection/lidc-idri/)
(**CC BY 3.0**). Slices resized/curated for GLIMPSE; originals at the Zenodo DOI.

**`ood/` — CT-ICH intracranial-hemorrhage scans** (Hssayeni et al., *Data* 2020),
**CC BY 4.0**, PhysioNet [ct-ich](https://physionet.org/content/ct-ich/),
DOI 10.13026/4nae-zg36.

> Tip: for clean license tagging you may prefer two separate HF dataset repos
> (LoDoPaB subset `odc-by`; CT-ICH OOD `cc-by-4.0`).

```bibtex
@article{leuschner2021lodopabct,
  title   = {LoDoPaB-CT, a benchmark dataset for low-dose computed tomography reconstruction},
  author  = {Leuschner, Johannes and Schmidt, Maximilian and Baguer, Daniel Otero and Maass, Peter},
  journal = {Scientific Data}, volume = {8}, number = {1}, pages = {109}, year = {2021}
}
@article{hssayeni2020ctich,
  title   = {Intracranial Hemorrhage Segmentation Using a Deep Convolutional Model},
  author  = {Hssayeni, Murtadha and Croock, Muayad and Salman, Aymen and Al-khafaji, Hassan and Yahya, Zakaria and Ghoraani, Behnaz},
  journal = {Data}, volume = {5}, number = {1}, pages = {14}, year = {2020}
}
```