Seminal breakthroughs in microscopy have repeatedly reshaped our understanding of biological systems by revealing previously unseen structures and dynamics. Today, we stand at the brink of another imaging revolution, driven by next-generation hardware innovations, petabyte-scale computational solutions, and novel sample-processing methods. In my talk, I will highlight our new adaptive optical multimodal microscope, which generates data at ~4 terabytes per hours and seamlessly integrates at least ten imaging modes, enabling rapid 5D (3D + chemistry + time) capture at nanometer–millimeter scales over millisecond–day timescales.
To handle these massive data streams, we developed PetaKit5D, a suite of high-performance computational tools that process images in real time, unlocking the full potential of modern light sheet microscopes. By offering memory-efficient data I/O, rapid geometric transformations, fast Richardson-Lucy deconvolution, and scalable Zarr-based stitching, PetaKit5D overcomes bottlenecks that previously rendered petabyte-scale acquisitions impractical. This synergy of hardware and software paves the way for peta-to-exabyte-scale imaging and analysis that was previously impractical.
Finally, I will introduce volumetric imaging via photochemical sectioning (VIPS) to overcome depth limitations in fluorescence microscopy using our next-generation hydrogel. VIPS is a light-driven technique, combined with volumetric lattice light-sheet imaging and petabyte-scale computational analysis, which enables true whole-mount nanoscale imaging from subcellular structures to entire organs.
The greatest remaining challenge is extracting biologically meaningful insights from these immensely large and complex datasets. If time permits, I will discuss our efforts and vision to build a 4D foundation model that empower scientists to harness large 4/5D datasets to accelerate the pace of scientific discovery.