Biologically inspired computer models of the microscopic and macroscopic structure of the brain based on wiring optimisation

Abstract

Like all natural systems, the mammalian brain and its neurons are governed by the fundamental principles of physics. Simplified computer simulations based on such principles help us see through the high complexity of especially the human brain and ultimately figure out how it works. In the scope of this thesis, two biologically realistic models were created focusing on, firstly the macroscopic and secondly the microscopic structure of the brain. The key component to both of these morphological simulations is the principle of wiring optimisation. First, combining dimensionality reduction methods and biologically inspired modelling based on optimal wiring, this thesis develops a method to simulate how the gyrification pattern of the mammalian brain emerges, differs from species to species, and changes due to pathological changes in neuron connectivity. The gyrification model is based on two biology-driven key principles: First, neuron placement follows wiring optimisation requirements and second, local connectivity between neurons is strong while long range connectivity is sparse as observed in the mammalian cortex. Many studies from the past saw the formation of gyri and sulci as the result of the surface of the cortex trying to expand in the limited cavity of the skull. The simulation described here shows that even without the constraint of the skull, gyrification still emerges when applying a biological neural connectivity distribution in addition to wiring optimisation. The first model provides new insights into the macroscopic structure of the brain but lacks microscopic detail. The second model is also based on optimal wiring but focused on reproducing the anatomical neuronal structure at the level of single cells. It provides a new algorithm and a tool to repair and preserve the microscopic structure of neuron morphology reconstructions. This is especially relevant for human neurons since here, data is extremely hard to come by, and the data that is available mostly originates from patients with diseases like severe epilepsy. Since the reconstruction process is a delicate procedure, the anatomical structure of reconstructed neurons is oftentimes severed by dendrites accidentally being cut. The recovered anatomy of neuronal dendrites is, however, pivotal to study the functionality of human and nonhuman neurons, which is further illustrated by analysing passive electrophysiological differences between human and mouse neurons. In summary, the thesis shows that optimal wiring is a useful guiding principle to simulate and better understand macroscopic and microscopic anatomical structure of the brain at the level of cortical folding as well as individual dendritic trees of nerve cells.