Metadata-Version: 2.4
Name: med-imagenet
Version: 0.1.0
Project-URL: Homepage, https://github.com/bhklab/med-imagenet
Project-URL: Source, https://github.com/bhklab/med-imagenet
Project-URL: Documentation, https://bhklab.github.io/med-imagenet/dev
Author-email: Joshua Siraj <joshua.siraj@uhn.ca>, Declan Korda <dkorda@protonmail.com>
License-Expression: MIT
License-File: LICENSE
Requires-Python: <3.14,>=3.10
Requires-Dist: click<9,>=8.2.1
Requires-Dist: fsspec<2027,>=2026.3.0
Requires-Dist: huggingface-hub<1,>=0.20
Requires-Dist: idc-index<0.12,>=0.11.10
Requires-Dist: med-imagetools<3,>=2.11.1
Requires-Dist: msgpack<2,>=1.1.2
Requires-Dist: orjson<4,>=3
Requires-Dist: pandas<3,>=2.2.3
Requires-Dist: platformdirs<5,>=4
Requires-Dist: pydantic<3,>=2.12.5
Requires-Dist: requests<3,>=2.32
Requires-Dist: rich<15,>=14.1.0
Requires-Dist: s3fs<2027,>=2026.2.0
Requires-Dist: structlog<26,>=25.5.0
Requires-Dist: tqdm<5,>=4.67.3
Description-Content-Type: text/markdown

# Med-ImageNet: A Standardized Resource for AI-Ready Oncology Imaging

## Core Features

Med-ImageNet is an open-source platform that transforms heterogeneous
cancer imaging collections into harmonized, AI-ready resources for
oncology research. It provides tools to **query**, **download**, and
**preprocess** medical imaging datasets from public and user-provided
sources through a unified Python interface.

![Index](docs/figures/overview.png)

## Platform Components

The platform comprises three integrated components:

1. **Med-ImageDB** -- Dataset indexing, query API, and secure image and
   metadata retrieval across all supported collections. The index can be found
   [here](https://huggingface.co/datasets/bhklab2026/med-image-index).

2. **Med-ImageTools** --
   Standardized preprocessing including DICOM ingestion, voxel harmonization,
   intensity normalization, and metadata alignment. The tools can be found [here](https://github.com/bhklab/med-imagetools).

3. **Med-ImageNet Repository** -- Unifies these modules into a scalable and
   reproducible data compendium supporting both raw data access and AI-ready
   outputs (e.g., NIfTI format) for deep learning integration.

![Architecture](docs/figures/architecture.png)

## Installing Med-ImageNet

```console
pip install med-imagenet
```

```console
imgnet --help
```

## Key Capabilities

- Queries across **all supported collections** with associated metadata
- Establishes explicit links between **paired imaging modalities**
  (e.g., CT with RTSTRUCTs)
- Query and request datasets based on **imaging region** and **imaging modality**
- Downloads from TCIA/IDC, S3, Dropbox, Zenodo, and HuggingFace sources
- Processes raw DICOM files to generate **AI-ready NIfTI outputs**,
  tabular metadata files, and dataset summaries

## License

This project uses the following license: [MIT License](https://github.com/bhklab/med-imagenet/blob/main/LICENSE)
