tiamat

Tiled Image Access, Manipulation, and Analysis Toolkit

tiamat is a modular Python toolkit for accessing, transforming, and exposing large scientific image datasets. It provides a flexible, pluggable pipeline model that separates data access (readers), transformation (transformers), and delivery (interfaces) — allowing on-the-fly, tool-agnostic image workflows without data duplication or format conversion.

Supported outputs include NumPy arrays, Napari, Neuroglancer, OpenSeadragon, and FUSE-mounted virtual filesystems.

📑 Table of Contents

1. Quick Start
2. Installation
3. Core Concepts
4. Examples
5. Development Guidelines
6. Contributing
7. Acknowledgements
8. License

🚀 Quick Start

from tiamat import Pipeline
from tiamat.io import ImageAccessor
from tiamat.transformers import FractionalTransformer, LUTTransformer

# Create a pipeline with fractional coordinate access and a rainbow colormap
pipeline = Pipeline(
    access_transformers=[FractionalTransformer()],
    image_transformers=[LUTTransformer(colormap="rainbow")]
)

# Request the central 50% of the image
accessor = ImageAccessor(x=(0.25, 0.75), y=(0.25, 0.75))
result = pipeline("example_image.tif", accessor=accessor)

# Get the transformed NumPy image and metadata
image = result.image
metadata = result.metadata

📦 Installation

Install the latest release from pypi:

pip install tiamat-python

Install the latest development version directly from GitLab:

pip install git+https://jugit.fz-juelich.de/inm-1/bda/software/data_access/tiamat/tiamat

🧠 Core Concepts

Tiamat defines a modular pipeline composed of:

• Readers: Load image data from formats like TIFF, NIfTI, HDF5, or memory arrays.
• Transformers: Apply dynamic, on-the-fly transformations (e.g., colormaps, axis reordering, tiling).
• Interfaces: Serve data to tools like Napari, Neuroglancer, OpenSeadragon, or directly as arrays.

This decoupled architecture allows you to:

• Build pipelines from reusable components
• Extend with custom readers or transformers
• Avoid costly format conversions

Interfaces

Interfaces use tiamat to expose data to various client applications.

• tiamat-openseadragon: Interface compatible with OpenSeadragon.
• tiamat-ng: Interface compatible with Neuroglancer.
• tiamat-fuse-zarr: Interface for exposing tiamat pipelines as zarr files through FUSE (experimental).
• tiamat-napari: Napari plugin interface for tiamat (experimental).

Extension transormers and readers

• tiamat-justice: Specialized readers and transformers used at INM-1, Forschungszentrum Jülich.
• tiamat-celldetection: AI-based transformers for live cell segmentation.

📁 Examples

See the examples/ directory for usage demonstrations and pipeline configurations.

🛠️ Development Guidelines

• Follow PEP 561 type hinting
• Use Google-style docstrings
• Formatting: flake8 with line length 120
• Tests: pytest unit tests
• Feature development follows git-flow

Releases

Releases to pypi are automatically performed on semantic versioning tags on the master branch.

🤝 Contributing

We welcome contributions!

• Fork the repository and work on a feature branch.
• Submit a Merge Request (MR) into develop.
• All contributions are reviewed and tested before merging.

This project follows the git-flow workflow. Releases are merged into master from develop on a regular basis.

📄 License

Apache 2.0 – see LICENSE for details.

🙏 Acknowledgements

See ACKNOWLEDGEMENTS.