Quantitative cell phenotyping across scales
We build computational frameworks that extract rich, quantitative phenotypes from time-lapse microscopy — capturing shape, appearance, and motion without requiring manual annotation. Our MOSES framework uses superpixel optical flow to discover collective migration phenotypes; SPOT/SAM provides a standardised vocabulary for comparing phenotypes across experiments and organisms. Our u-segment3D tool achieves universal 3D cell segmentation by combining diverse 2D models, removing the need to retrain for each imaging modality.
Next-generation fluorescence imaging systems
We collaborate on the development and application of advanced light-sheet imaging platforms — including oblique plane microscopy (OPM), structured illumination OPM, projective light-sheet, and self-driving multiscale microscopy — that image live specimens from whole-organism to subcellular resolution. We also develop cyclic immunofluorescence (CyCIF) in 3D to enable highly multiplexed protein profiling in intact tissue volumes, and apply expansion microscopy for nanoscale subcellular mapping.
Morphology-driven oncogenic signalling hubs
We investigate how cell morphology controls signal transduction. Our work showed that plasma-membrane blebs act as compartments that concentrate and activate pro-survival oncogenic complexes (Nature 2023). We continue to dissect how branched actin-driven protrusions regulate KRAS signalling, and how amoeboid migration modes allow cancer cells to invade tissues without matrix proteolysis. These studies combine live-cell super-resolution imaging, optogenetics, and quantitative signal analysis.
Surface-guided computing for subcellular morphodynamics in tissue context
We develop surface-guided computational methods that map molecular signals onto the 3D cell surface to study how signal distributions co-vary with shape dynamics. Combined with volumetric cyclic immunofluorescence and universal 3D segmentation (u-segment3D, Nature Methods 2025), these approaches allow us to profile cell states and subcellular organisation in intact tumour and embryonic tissues at single-cell and subcellular resolution, linking morphodynamic phenotypes to molecular identity and tissue-level context.