TileDB-BioImaging

Python package for:

• converting images stored in popular Biomedical Imaging formats to groups of TileDB arrays (& vice versa)
• exposing an expressive and efficient API (powered by TileDB) for querying such data.

Note: this project is in pretty early stage and under heavy development. Breaking changes to the underlying data format and the API are to be expected.

Features

Ingestion to TileDB Groups of Arrays

- OME-Zarr
- OME-Tiff
- Open-Slide

Export from TileDB-Bioimaging Arrays to:

- OME-Zarr
- OME-Tiff

Visualization Options

• TileDB Cloud includes a built-in, pyramidal multi-resolution viewer (log in to TileDB Cloud to see an example image preview here)
• Napari: https://github.com/TileDB-Inc/napari-tiledb-bioimg

Quick Installation

• From PyPI:

  # If you use zsh replace [full] with \[full\]
  pip install tiledb-bioimg[full]
  

• From source:

  git clone https://github.com/TileDB-Inc/TileDB-BioImaging.git
  cd TileDB-BioImaging
  
  # If you use zsh replace .[full] with .\[full\]
  pip install -e .[full]
  

Note: For tiff support ([tiff] or [full] extra dependencies) on Python 3.7, you should first install our tifffile fork rather than the older PyPI upstream package:

  pip install tifffile@git+https://github.com/TileDB-Inc/tifffile.git@gsa/python-3.7

Examples

How to convert imaging data from standard biomedical formats to group of TileDB arrays.

OME-Zarr to TileDB Group of Arrays

from tiledb.bioimg.converters.ome_zarr import OMEZarrConverter
OMEZarrConverter.to_tiledb("path_to_ome_zarr_image", "tiledb_array_group_path")

OME-Tiff to TileDB Group of Arrays

from tiledb.bioimg.converters.ome_tiff import OMETiffConverter
OMETiffConverter.to_tiledb("path_to_ome_tiff_image", "tiledb_array_group_path")

Open Slide to TileDB Group of Arrays

from tiledb.bioimg.converters.openslide import OpenSlideConverter
OpenSlideConverter.to_tiledb("path_to_open_slide_image", "tiledb_array_group_path")

Documentation

API Documentation is auto-generated. Following the instructions below:

cd docs && make html

will generate a docs/_build directory, where you can now open the index.html