Advanced in vitro models (e.g., organoids) are three-dimensional (3D) constructs usually generated from cells with a degree of stemness. This, accompanied by the multiplicity of parameters which condition organoid growth and morphology, results in constructs of different shapes and size and thereby functional properties. To date, a quantitative framework for robust measurement and modelling of organoid growth processes in relation to morphological features is still lacking and urgently required. In silico methods integrate physical data and biochemical models using computational tools, and are a powerful support for tissue engineering. They are particularly relevant for studying organoids: the multiplicity of parameters which condition their growth and morphology can be explored in virtual models, facilitating experimental design, and enabling prediction and extrapolation of behaviour and function. Here we describe a framework for quantifying organoid morphometry with imaging tools and mapping shape and size to virtual organoids generated through evolutionary algorithms.
Multi tissue hepatic organoids and spheroids are generated using standard protocols. They are imaged at multi-scale by means of an integrated approach involving light-sheet, confocal and super-resolution microscopy. Thus, we can resolve objects spanning from few millimetres (e.g., construct shape) down to microns (e.g., cells) to tens of nanometers (e.g., mitochondria, tight junctions). Ad hoc image processing algorithms and routines are developed for dealing with the acquired datasets. In particular, algorithms based on the intensity distributions of background and foreground, described locally within the dataseta, are exploited for identifying and isolate single cells and/or subcellular constructs.
Evolutionary algorithms based on the optimization of a cost function which incorporates resource uptake, surface energy and cooperative metabolic effort are used to generate virtual organoids within a range of masses.
Quantitative descriptors to characterize construct and cell shape, cell arrangement in the 3D space as well as cell-cell and cell-substrate interactions are identified. Evolutionary algorithms are honed by matching imaging data with the virtual organoids. We are thus able to identify how resource assimilation and physical phenomena affect organoid formation and growth.
Our framework enables the characterisation of structural features of 3D constructs, which in turn may give insights on their functionality. In addition, the outputs are used for implementing more accurate computational models that take into account the real shape of the constructs and the real arrangement of the cells. This will serve to quantitatively assess to what extent organoids are similar to their in vivo counterparts, and thus define strategies for improving reproducibility and viability. Integrating experimental and modelling approaches is key for designing constructs with translational value and hence useful for robust in vitro to in vivo extrapolation, paving the way towards predictive and precision medicine and reducing animal tests. Currently the models are deterministic, future efforts will be dedicated to incorporating fluctuations as an inevitable and ubiquitous feature of any functional biological system.