Xiongtao Ruan

Understanding biological systems requires observing features and processes across vast spatial and temporal scales, spanning nanometers to centimeters and milliseconds to days, often using multiple imaging modalities within complex native microenvironments. Yet, achieving this comprehensive view is challenging because microscopes optimized for specific tasks typically lack versatility due to inherent optical and sample handling trade-offs, and frequently suffer performance degradation from sample-induced optical aberrations in multicellular contexts. Here, we present MOSAIC, a reconfigurable microscope that integrates multiple advanced imaging techniques including light-sheet, label-free, super-resolution, and multi-photon, all equipped with adaptive optics. MOSAIC enables non-invasive imaging of subcellular dynamics in both cultured cells and live multicellular organisms, nanoscale mapping of molecular architectures across millimeter-scale expanded tissues, and structural/functional neural imaging within live mice. MOSAIC facilitates correlative studies across biological scales within the same specimen, providing an integrated platform for broad biological investigation.

Optical nanoscopy of intact biological specimens has been transformed by recent advancements in hydrogel-based tissue clearing and expansion, enabling the imaging of cellular and subcellular structures with molecular contrast. However, existing high-resolution fluorescence microscopes have limited imaging depth, which prevents the study of whole-mount specimens without physical sectioning. To address this challenge, we developed "photochemical sectioning," a spatially precise, light-based sample sectioning process. By combining photochemical sectioning with volumetric lattice light-sheet imaging and petabyte-scale computation, we imaged and reconstructed axons and myelination sheaths across entire mouse olfactory bulbs at nanoscale resolution. An olfactory-bulb-wide analysis of myelinated and unmyelinated axons revealed distinctive patterns of axon degeneration and de-/dysmyelination in the neurodegenerative mouse, highlighting the potential for peta- to exabyte-scale super-resolution studies using this approach.

Light sheet microscopy is a powerful technique for high-speed 3D imaging of subcellular dynamics and large biological specimens. However, it often generates datasets ranging from hundreds of gigabytes to petabytes in size for a single experiment. Conventional computational tools process such images far slower than the time to acquire them and often fail outright due to memory limitations. To address these challenges, we present PetaKit5D, a scalable software solution for efficient petabyte-scale light sheet image processing. This software incorporates a suite of commonly used processing tools that are memory and performance-optimized. Notable advancements include rapid image readers and writers, fast and memory-efficient geometric transformations, high-performance Richardson-Lucy deconvolution, and scalable Zarr-based stitching. These features outperform state-of-the-art methods by over one order of magnitude, enabling the processing of petabyte-scale image data at the full teravoxel rates of modern imaging cameras. The software opens new avenues for biological discoveries through large-scale imaging experiments.

Lattice light sheet microscopy excels at the noninvasive imaging of three-dimensional (3D) dynamic processes at high spatiotemporal resolution within cells and developing embryos. Recently, several papers have called into question the performance of lattice light sheets relative to the Gaussian sheets most common in light sheet microscopy. Here, we undertake a theoretical and experimental analysis of various forms of light sheet microscopy, which demonstrates and explains why lattice light sheets provide substantial improvements in resolution and photobleaching reduction. The analysis provides a procedure to select the correct light sheet for a desired experiment and specifies the processing that maximizes the use of all fluorescence generated within the light sheet excitation envelope for optimal resolution while minimizing image artifacts and photodamage. We also introduce a new type of “harmonic balanced” lattice light sheet that improves performance at all spatial frequencies within its 3D resolution limits and maintains this performance over lengthened propagation distances allowing for expanded fields of view. Comparing conventional and lattice light sheets reveals the advantages of the latter and how to improve their performance further.

Cell shape provides both geometry for, and a reflection of, cell function. Numerous methods for describing and modeling cell shape have been described, but previous evaluation of these methods in terms of the accuracy of generative models has been limited. Here we compare traditional methods and deep autoencoders to build generative models for cell shapes in terms of the accuracy with which shapes can be reconstructed from models. We evaluated the methods on different collections of 2D and 3D cell images, and found that none of the methods gave accurate reconstructions using low dimensional encodings. As expected, much higher accuracies were observed using high dimensional encodings, with outline-based methods significantly outperforming image-based autoencoders. The latter tended to encode all cells as having smooth shapes, even for high dimensions. For complex 3D cell shapes, we developed a significant improvement of a method based on the spherical harmonic transform that performs significantly better than other methods. We obtained similar results for the joint modeling of cell and nuclear shape. Finally, we evaluated the modeling of shape dynamics by interpolation in the shape space. We found that our modified method provided lower deformation energies along linear interpolation paths than other methods. This allows practical shape evolution in high dimensional shape spaces. We conclude that our improved spherical harmonic based methods are preferable for cell and nuclear shape modeling, providing better representations, higher computational efficiency and requiring fewer training images than deep learning methods.