Definiens Health Image Intelligence - From Images to Clinical Decisions - Measuring the Spatial Relationships of Endosomes in Hepatocytes

In systems biology, knowledge is generated through data analysis, mathematical modelling, simulation, and in-vitro and in-vivo experiments. In this overall process, image data is increasingly establishing itself as the information carrier between experiment, modelling, and simulation. Analyzing the image data provides important parameters for both the models and the experiments. The parameters generated from the image data can then serve as input data for the experiments and their accompanying models and simulations.

In the present study – which is part of a large-scale research framework to investigate the multitude of interactions during endocytosis from a systems biology perspective – confocal stacks of hepatocytes were analyzed using Definiens Developer XD to measure the spatial distribution of vesicles, so-called endosomes.

Pixels of the native 3D image are grouped to create objects, which form cell compartments like nuclei, endosomes, and cytoplasm, which in turn form hepatocytes. Analysis of the image data can occur in different modalities and resolutions, and image metadata can also be referenced in order to supplement the analysis process. It is feasible to analyze the simulation results from biological systems in a similar manner to experimental results, such as endocytosis simulations focussing on intracellular vesicle and particle interaction. Particle coordinates from the simulation results are transformed into a time series of 3D image stacks. Each image stack is then analyzed so as to identify vesicles from the collections of sub-cellular particles – grouped as particles that are either membrane- or cluster-forming (Rab5 Proteins, lipids, etc.), or grouped as cytosol-forming particles.

A series of additional transformations are performed to further divide the vesicles into particle-dense and non-particle-dense regions. This provides a segmentation framework for future time series analysis to identify interaction events between vesicles, such as budding, fusing, etc. Sub-regions with high particle densities are typically indicative of ongoing vesicle interaction, with particles being exchanged between vesicles. As information from both experimental and simulation measurements can be processed down to a shared level, this information can also be transformed from one domain to another. The creation and improvement of simulation models derived from experimental data – or vice versa of experimental interpretation models derived from simulation data – is strongly supported and objectified through automated image analysis.

: Image analysis of simulation results. Left: Classification and object tracking of vesicles. Each image represents one time step in the simulation. Each object is assigned its own class (color-coded). Right: An automatically generated graph shows events of vesicle interactions. Simulation results: J. McCaskill Group, Ruhr University Bochum, Germany

Measuring the Spatial Relationships of Endosomes in Hepatocytes:|Mediation between Experiment, Simulation and Model

About Definiens