Supporting Tools for Topological Image Analysis

In bioimaging, the spatial relationships between entities such as cells are of great interest. For example, understanding how interactions between tumor cells and immune cells shape the microenvironment of a tumor may lead to novel approaches in cancer therapy.

This project will develop a napari plugin for interactive construction and visualization of spatial object graphs from pairs of images and segmentation masks. To this end, the team will introduce a new graph layer type, which can be overlayed on loaded images and segmentation masks. At their core, such graph layers hold networkx or igraph objects (backend to be decided), integrating popular Python network analysis tools into napari. Implementing napari’s IO hook specification, graph layers can be saved to/loaded from standard graph file formats supported by a wide range of third-party software. For interactive data analysis, the Python graph objects are exposed to the user. Building on the framework for multi-channel image processing, steinbock, the plugin provides a dock widget for interactively constructing spatial object graphs from loaded pairs of images and segmentation masks with user-defined parameters.

The group will combine graph construction implemented in steinbock with popular Python packages for network analysis to enable spatial object graph visualization and topological image analysis from within napari.