In-silico study on the dynamical and topological relationships between three-dimensional cultures and relative slices

Three-dimensional (3D) in-vitro neuronal cultures represent a pioneering technological advance in exploring dynamics, function and dysfunction in environments that mimic brain's rich spatial organization. However, the recording of activity on the entire 3D network remains a challenge as researchers often resort to methods developed for two-dimensional (2D) cultures, such as multi-electrode arrays (MEAs) or calcium fluorescence imaging on a just a focal plane. Three-dimensional data acquisition methods such as light-sheet fluorescence microscopy (LSFM) require specialized facilities and advanced data processing. In either MEA or single calcium imaging plane approaches, the question of whether a 2D readout layer reliably captures topological and dynamical properties of the entire 3D network quickly arises. This question is addressed in the present work by means of in-silico simulations of a 3D culture analysed along different developmental stages. To mimic the single-plane readout of a typical 2D experimental design, the bottom layer (slice) of the 3D neuronal assembly was drawn out and its features were compared with the entire 3D parent network. The results show that, in terms of qualitative trends, the slice successfully captures the features of the 3D network. However, the amplitude of network-wide events and its variability were systematically overestimated by the slice, despite accurately capturing their duration and frequency. Additionally, some network metrics such as small-worldness were underestimated, while others such as modularity were regularly overestimated. Our study is a preliminary investigation that aims to provide insight into the relationship between the whole 3D network's properties and the features extracted from just a recorded read-out subnetwork, with the aim of better understanding experimental data and its limitations in interpretation.