To realise biological function, cellular behaviour is a dynamic, involved processes that encompass spatial arrangement, differentiation, division and death. Numerous in vivo biological experiments have been designed to track cellular behaviour over different timeand space-scale frames, generating a large amount of spatio-temporal data. However, traditional statistical methods are not well adapted to draw meaningful insight from those high-dimensional data. Motivated by primary data sources provided by collaborators, this thesis presents a novel statistical analysis framework of high-dimensional spatio-temporal data, addressing it at different time scales. To capture biological information at fixed time points, a spatial statistical pipeline is developed to quantify the distribution of various cell types and assess their spatial relationships. The analysis is extended to cover periods of the order of one day, with a statistical framework designed to process spatio-temporal data, combining a data cleaning process and investigating the relationship between cell movement and differentiation. Over longer time periods, mathematical models and statistical methods are developed to estimate the average number of divisions in vivo, offering insights into long-term cell distribution. The advanced statistical analysis helps capture the spatio-temporal relationships between different cell types, revealing the dynamic processes of cellular behaviour.