Fig. 2.

Investigating tissue hydraulics in mammalian ovarian folliculogenesis using an integrative approach that combines advanced biophotonics, deep learning and biophysical modelling. A Label-free RI imaging combined with deep learning enables visualisation of subcellular structures and dynamics with high specificity (Jo et al. 2021). B Deep learning model detects cell divisions during wound healing with high accuracy (Turley et al. 2024a). C 3D segmentation of GCs in the ovarian follicle. D Machine learning based on physics-informed neural networks allows inference of traction forces in a cell (Schmitt et al. 2024). E Computational approach to model luminogenesis in follicles based on fluid mixing-demixing transition. F Measurement of the surface tension of ovarian follicle using micropipette aspiration. G Representative “stiffness map” of an ovarian follicle imaged by Brillouin microscopy, revealing clear intrafollicular mechanical heterogeneities (Chan et al. 2021). H An image of pre-ovulatory follicle acquired with OCM. Scale bar, 100 $\mu$m. I Images showing a cell undergoing division, acquired through QPI. Interstitial fluids are also visible due to its distinct RI from cellular bodies