Internal KCN: Biophysically Constrained Generative Algorithm Synthesizing Realistic Astrocytes in silico

Astrocytes are prominent glial cells in the brain that contact multiple synapses and form lateral connections between them, where the activity at one synapse may modulate another through internal pathways of the astrocyte. To understand this connection, it is essential to characterize and reproduce astrocyte branching structures in 3D. However, such a framework does not exist currently. We extract essential morphological features from experimental 3D astrocyte tracings. Deploying techniques from statistics, machine learning, and graph theory, we develop a generative algorithm to grow artificial astrocyte branching structures inspired by biophysical constraints. The generated branching structures reproduce essential morphological features observed in experimental data. Therefore, our algorithms fill in the current gap of astrocyte anatomy characterization, and the inner workings of the generative process shed light on the potential biophysical forces that shape the astrocyte branching structure.