Interactive visualization of bio-imaging data

It is recommended to install this repository in a dedicated conda environment:

conda create -n viz python=3.7
conda activate viz
conda install -c pyviz holoviews bokeh panel param xarray datashader pytables

git clone https://github.com/fmi-basel/inter-view.git
pip install inter-view/

Notebooks for existing use cases are available under notebooks/applications. These applications can be directly used as a notebook or, once configured, rendered in a browser with:

panel serve --show PATH_TO_NOTEBOOK.ipynb

or remotely:

panel serve --port PORT --allow-websocket-origin WORKSTATION:PORT PATH_TO_NOTEBOOK.ipynb

TIF_OVR viewer: Browsing through large TIF overviews and exporting RGB images.

Segmentation viewer: Inspecting segmentation labels and exporting mispredicted samples to retrain a deep learning network.

2D annotator: Tool to correct annotations and rapidly switch between samples. It also works directly on large (>10 000 px) images.

3D annotator: Tool to correct annotations of 3D stacks.

Linked scatter plot: Visualization of extracted features (e.g. area, eccentricity, mean intensity, etc.) with a link to the original images.

Examples of the available modules are provided in notebooks/module_examples. This is also intended as a makeshift unit test of the gui.

The repository is structured in sub-modules to load (io), view (view_images), and edit (edit_images) images. While these parts should be relatively modular/reusable, the dashboards sub-modules derives more complex applications that might sometimes be difficult to combine/extend.