OptiCell3D: Precise inference of mechanical cell properties from microscopy imaging
Runser, S.; Yamauchi, K. A.; Almanstoetter, M.; Lampart, F. L.; Carrara, F.; Conrad, L.; Schaumann, L.; Vetter, R.; Iber, D.
Show abstract
We introduce OptiCell3D, an open-source image-based framework for inferring cellular mechanical properties directly from 3D microscopy data. By integrating short physical simulations with gradient-based optimization via backpropagation, our method substantially improves accuracy compared to existing approaches. Moreover, OptiCell3D enables mechanical inference in tissues that are too large to be fully imaged, extending its applicability to more complex biological systems. We demonstrate the power of our approach by applying OptiCell3D to five morphologically diverse mouse epithelial tissues. By combining inferred parameters with simulations and morphometric analysis, we find that the ratio of apical to lateral surface tension predicts cell aspect ratio across epithelial subtypes, linking a single mechanical parameter to the broad morphological diversity of epithelia. Finally, we apply our framework to stratified tissues, finding greater variability in pressure and surface tension between cells and tension gradients along the apico-basal axis.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.