ACS Photonics

Localization microscopy techniques have overcome the diffraction limit, enabling nanoscale biological imaging by precisely determining the positions of individual emitters. However, the performance of deep learning methods commonly applied to these tasks often depends significantly on the quality of training data, typically generated through simulation. Creating simulations that perfectly replicate experimental conditions remains challenging, resulting in a persistent simulation-to-experiment (sim2exp) gap. To bridge this gap, we propose a physics-informed generative model leveraging self-supervised learning directly on experimental data. Our model extends the Deep Latent Particles (DLP) framework by incorporating a physical Point Spread Function (PSF) model into the decoder, enabling it to disentangle learned realistic environments from precise emitter properties. Trained directly on unlabeled experimental images, our model intrinsically captures realistic background, noise patterns, and emitter characteristics. The decoder thus acts as a high-fidelity generator, producing fully labeled, realistic training images with known emitter locations. Using these generated datasets significantly improves the performance of supervised localization algorithms, particularly in challenging scenarios such as complex backgrounds and low signal-to-noise ratios. Our results demonstrate substantial improvements in localization accuracy and emitter detection, underscoring the practical benefit of our approach for real-world microscopy applications. We will make our code publicly available.

Fluorescence Lifetime Imaging Microscopy (FLIM) quantifies the autofluorescence lifetime to measure cellular metabolism, therapeutic efficacy, and disease progression. These dynamic processes are intrinsically heterogeneous, increasing the complexity of the signal analysis. Often noise reduction strategies that combine thresholding and non-selective data smoothing filters are applied. These can result in error introduction and data loss. To mitigate these issues, we develop noise-corrected principal component analysis (NC-PCA). This approach isolates the signal of interest by selectively identifying and removing the noise. To validate NC-PCA, a secondary analysis of FLIM images of patient-derived colorectal cancer organoids exposed to a range of therapeutics was performed. First, we demonstrate that NC-PCA decreases the uncertainty up to 4-fold in comparison to conventional analysis with no data loss. Then, using a merged data set, we show that NC-PCA, unlike conventional methods, identifies multiple metabolic states. Thus, NC-PCA provides an enabling tool to advance FLIM analysis across fields.

Obtaining structural information from the enteric nervous system (ENS) within intact intestinal tissue requires microscopy systems capable of imaging through multiple tissue layers and during ongoing physiological motion. Tissue opacity, three-dimensional geometry, and spontaneous contractions strongly constrain volumetric imaging, limiting the applicability of most conventional linear optical techniques to imaging in either dissected, stretched or pharmacologically suppressed tissues. We apply Spectro-temporal Laser Imaging by Diffracted Excitation (SLIDE) microscopy, a diffraction-based scanning approach enabling fast volumetric two-photon imaging, to record the ENS in an intact ex vivo intestinal preparation from a transgenic mouse line expressing the red fluorescent protein TdTomato in peripheral and enteric neurons and glia. We achieved fast volumetric imaging during spontaneous contractions, capable of resolving micrometer-scale displacements in three dimensions, without inducing observable photodamage or compromising tissue viability over the experimental timescale. This work establishes 4D-SLIDE microscopy as a robust experimental framework for visualizing enteric neural structures within their native three-dimensional context during physiological motion, with direct relevance for conditions involving altered intestinal mechanics.

Imaging Fluorescence Correlation Spectroscopy (Imaging FCS) is a powerful tool to extract information on molecular mobilities, actions and interactions in live cells, tissues and organisms. Nevertheless, several limitations restrict its applicability. First, FCS is data hungry, requiring 50,000 frames at 1 ms time resolution to obtain accurate parameter estimates. Second, the data size makes evaluation slow. Thirdly, as FCS evaluation is model-dependent, data evaluation is significantly slowed unless analytic models are available. Here we introduce two convolutional neural networks (CNNs) - FCSNet and Im-FCSNet - for correlation and intensity trace analysis, respectively. FCSNet robustly predicts parameters in 2D and 3D live samples. ImFCSNet reduces the amount of data required for accurate parameter retrieval by at least one order of magnitude and makes correct estimates even in moderately defocused samples. Both CNNs are trained on simulated data, are model-agnostic, and allow autonomous, real-time evaluation of Imaging FCS measurements.

Optical scattering in biological tissues fundamentally limits fluorescence imaging by disrupting spatial and angular information, thereby restricting volumetric visualization. Although hardware-intensive and computational approaches have advanced scattering microscopy, practical three-dimensional imaging through tissue remains constrained by instrumental complexity and axial ambiguity. Here, we present volumetric scattering microscopy (VSM), a scan-free, optical-computational framework for three-dimensional fluorescence imaging in scattering biological media. VSM captures angularly resolved speckle-encoded fluorescence using an aperture-segmented Fourier light-field configuration and reconstructs volumetric structure through adaptive feature-based descattering and joint sub-pupil alignment. This hybrid strategy preserves angular information embedded in scattered light without wavefront measurement or mechanical scanning, while maintaining the simplicity of a standard epi-fluorescence architecture. We demonstrate high-fidelity volumetric reconstruction across phantoms, engineered cellular systems, ex vivo tissues with volumetric muscle loss, and intact Xenopus embryos, achieving preserved spatial resolution, enhanced optical sectioning, and quantitative accuracy under strong scattering conditions. VSM supports large-field, robust volumetric imaging in both layered and fully embedded scattering environments. By transforming scattered light into a structured encoding resource, VSM establishes a scalable pathway toward routine three-dimensional fluorescence imaging in complex biological systems.

Imaging subcellular structures deep within thick, turbid biological tissues remains fundamentally limited by light scattering, which distorts optical wavefronts and degrades contrast, resolution, and sensitivity. These limitations hinder quantitative interrogation of complex biological systems where resolving dynamic microenvironments at subcellular resolution is critical. Here, we introduce scattering-enabled epi-quantitative phase imaging (SEEQPI), a label-free method that leverages tissue scattering and provides subcellular spatial resolution, nanometer-scale spatiotemporal phase sensitivity, and millimeter-scale imaging depth in murine brains. SEEQPI is enabled by common-path phase-shifting confocal epi-interferometry with near-infrared illumination and the scattering-enabled phase reconstruction algorithm. SEEQPI requires low illumination power, minimizing tissue damage while enabling high-speed imaging of biological dynamics. We demonstrate simultaneous, colocalized imaging of subcellular structures with SEEQPI, third-harmonic generation, and three-photon fluorescence microscopy in liver cancer spheroids and in vivo mouse brains. SEEQPI enables quantitative, longitudinal studies of dry mass dynamics in intact, living biological systems.

Fluorescence Lifetime Imaging Microscopy (FLIM) enables quantitative mapping of molecular environments in living systems with high biochemical specificity. However, spatial overlap dictated by the diffraction-limited point spread function (PSF) causes a mixing of temporal signals: photons from neighboring emitters collected within the same pixel yield composite decay profiles, generating apparent intermediate lifetimes that can be mistaken for variations in the local molecular environment. We introduce a workflow that applies Mean-Shift Super-Resolution (MSSR) to raw intensity data to generate intensity-derived spatial masks prior to phasor-based lifetime analysis. The method is computationally efficient and preserves decay kinetics because it operates on intensity-derived spatial information rather than modifying temporal data. In U2OS cells labeled with spectrally-overlapping fluorophores, phasor analysis reveals an intermediate lifetime population localized at PSF-overlap interfaces, consistent with optical mixing rather than intrinsic lifetime heterogeneity. MSSR-derived masking suppressed this mixed population while preserving stable phasor cluster centers -i.e. the distribution of similar phasor coordinates in the phasor plane- for each fluorophore. Simulations of strictly monoexponential fluorescence decay emitters further show that blended lifetime decay profiles are present at separations up to 4{sigma} and becomes maximal near [~]1.6{sigma}, indicating that conventional spatial resolution criteria can underestimate lifetime cross-talk. Application of this workflow to three-component FLIM showed also a reduced overlap of pixel distributions in phasor plots while maintaining distinct lifetime signatures. Overall, MSSR-based spatial refinement provides an accessible strategy to improve the spatial resolution while maintaining accuracy of FLIM measurements.

Super-resolution microscopy, particularly localization-based methods, necessitates careful balancing of optical complexity, computational demands, and user accessibility. Conventional strategies typically adopt either deterministic or learning-based approaches, overlooking opportunities to leverage their syn-ergistic strengths. In this work, we introduce deep volumetric localization microscopy (VLM), a super-resolution methodology that integrates instrumental and algorithmic advancements for high-fidelity 3D single-molecule imaging. VLM employs a wavefront-optimized light-field configuration to capture single-molecule data, while a cascaded neural network reconstructs 3D volumes and extracts molecular coordinates at a 10 nm and 25 nm localization precision in the lateral and axial dimensions, respectively, across an imaging depth exceeding 5 {micro}m. Unlike existing methods, VLM is trained exclusively with system-aware intrinsic point-spread functions, bypassing dependencies on external imaging modalities or sample-specific data training. We validate VLM across diverse biological specimens, demonstrating hardware simplicity, data efficiency, and minimal phototoxicity. We anticipate VLM will overcome current limitations in fluorescence microscopy, empowering broader advancements in biomedical research.

The application of genetically encoded fluorophores for microscopy has afforded one of the biggest revolutions in the biosciences. Bioluminescence microscopy is an appealing alternative to fluorescence microscopy, because it does not depend on external illumination, and consequently does neither produce spurious background autofluorescence, nor perturb intrinsically photosensitive processes in living cells and animals. The low quantum yield of known luciferases, however, limit the acquisition of high signal-noise images of fast biological dynamics. To increase the versatility of bioluminescence microscopy, we present an improved low-light microscope in combination with deep learning methods to increase the signal to noise ratio in extremely photon-starved samples at millisecond exposures for timelapse and volumetric imaging. We apply our method to image subcellular dynamics in mouse embryonic stem cells, the epithelial morphology during zebrafish development, and DAF-16 FoxO transcription factor shuttling from the cytoplasm to the nucleus under external stress. Finally, we concatenate neural networks for denoising and light-field deconvolution to resolve intracellular calcium dynamics in three dimensions of freely moving Caenorhabditis elegans with millisecond exposure times. This technology is cost-effective and has the potential to replace standard optical microscopy where external illumination is prohibitive.

Optical aberrations hinder fluorescence microscopy of thick samples, reducing image signal, contrast, and resolution. Here we introduce a deep learning-based strategy for aberration compensation, improving image quality without slowing image acquisition, applying additional dose, or introducing more optics into the imaging path. Our method (i) introduces synthetic aberrations to images acquired on the shallow side of image stacks, making them resemble those acquired deeper into the volume and (ii) trains neural networks to reverse the effect of these aberrations. We use simulations and experiments to show that applying the trained de-aberration networks outperforms alternative methods, providing restoration on par with adaptive optics techniques; and subsequently apply the networks to diverse datasets captured with confocal, light-sheet, multi-photon, and super-resolution microscopy. In all cases, the improved quality of the restored data facilitates qualitative image inspection and improves downstream image quantitation, including orientational analysis of blood vessels in mouse tissue and improved membrane and nuclear segmentation in C. elegans embryos.

Bessel beam has long been utilized in physics for its ability to maintain lateral confinement during propagation. When used for two-photon fluorescence microscopy, Bessel foci have enabled high-speed volumetric imaging of the brain. At high numeric aperture (NA), however, the substantial energy in the side rings of Bessel foci reduces image contrast. Therefore, a compromise between resolution and contrast has to be made, limiting Bessel foci in microscopy to low NA. Here, we describe a method of generating axially extended Bessel-droplet foci with much suppressed side rings. Shaping the excitation wavefront with novel phase patterns, we generated Bessel-droplet foci of variable NAs at high power throughput and scanned them interferometrically along the axial direction for continuous volume imaging. More resistant to optical aberrations than Bessel foci, Bessel-droplet foci enabled high-resolution and high-contrast volumetric imaging of synaptic anatomy and function as well as lymphatic circulation in the mouse brain in vivo.

The eye offers a unique non-invasive window for accessing single-cell level structures and functions of the central nervous system (CNS) throughout the retina. However, strong and space-varying ocular aberrations, along with limited volume rates, challenge large-scale cellular imaging in living eyes and stymie the full potential of possible biological and pathological studies in retina. Here, we present plenoptic illumination scanning laser ophthalmoscopy (PI-SLO), a 3D fluorescent retinal imaging modality that enables high-speed, widefield, volumetric single-cell imaging with low phototoxicity. By capturing multiple angular images of fluorescence signals from the entire volume, PI-SLO enables digital aberration correction and 3D imaging across a >20{o} FOV with >23 Hz volume rate. We leverage this structural and functional imaging modality to investigate three key aspects of CNS physiology through the living mouse retina, including: microglial process dynamics, vascular perfusion, and light evoked calcium fluxes in inner retinal neurons. PI-SLO is a versatile non-invasive platform for in vivo investigation of retinal and CNS physiology at the cellular level.

Single-molecule localization microscopy achieves nanometer-scale resolution but fails to count molecular targets when multiple targets are in close proximity. DNA-PAINT uses reversible binding of fluorescently labeled probes to image molecular targets, but existing time-series based counting methods require prior knowledge of binding kinetics, calibration, or extensive data post-processing, which limits applicability across heterogeneous biological samples. Here, we present mCOAST, a calibration-free method that extracts molecular counts and kinetic parameters directly from a single DNA-PAINT intensity trace using cumulants. Unlike existing approaches, mCOAST requires no adjustable parameters, data normalization, or denoising. We demonstrate accurate counting for diffraction-limited clusters with up to 48 targets at high imager concentrations, with precision that improves, rather than decreases, as concentration increases. Critically, mCOAST counts targets in individual clusters despite kinetic heterogeneity across samples or between experiments. This paves the way towards quantitative imaging and counting in uncalibrated biological systems, such as living cells.

Fluorescence microscopy is a cornerstone of biological research. However, fluorescent labeling is challenging in live cells and is constrained by photobleaching and phototoxicity. Label-free methods allow cells to be studied in their native state, but most techniques have poor contrast, lack 3D capability, rely on complex optics, and fail to provide structural information. We present broadband backscattering confocal microscopy (BBCM), which employs a broadband supercontinuum laser and collects backscattered light in confocal geometry using a photomultiplier tube. Broadband illumination averages out size-dependent oscillations that confound monochromatic backscattering. This eliminates blind spots and intensity ambiguities, allowing all scatterers to be visible, with the signal increasing approximately linearly with scatterer size. BBCM is easy to retrofit to standard confocal microscopes, requires no specialized optics, and is straightforward for nonspecialists. It enables high-contrast, label-free 3D imaging of live cells with size sensitivity to subcellular structures without employing custom optics or complex data processing.

Mobile biological particles, ranging from biomolecules to viral capsids, often diffuse faster than 1 {micro}m2/s, resulting in severe motion blur in conventional millisecond-scale imaging. While shorter exposures may help provide the data needed to capture faster dynamics, quantization of signal intensity per pixel at such exposures eventually interferes with our ability to track. In the extreme case of binary (1-bit-per-pixel) output-- where going from 8-bit conventional grayscale imaging to 1-bit directly corresponds to a 255-fold faster acquisition rate--no existing tracking methods can be used, as these methods fundamentally rely on intensity-based localization, which does not leverage the binary output. For this reason, we introduce single-photon single-particle tracking (SP2T), a framework that bypasses localization and linking by estimating particle numbers and trajectories directly by jointly considering 1-bit image stacks. While cameras capable of microsecond-scale exposures, typically based on scientific CMOS (sCMOS) sensors or single-photon detectors (SPDs), are increasingly central to this effort, in this work, we focus on single-photon avalanche diode (SPAD). Single-photon detector (SPD) arrays offer microsecond exposures over large fields of view (512x512 pixels). SP2T accounts for detector-specific artifacts such as hot and cold pixels and is validated with programmed fluorescent bead trajectories and biological systems (aerolysin and ganglioside). These experiments, in addition to simulations, reveal that analysis performed with longer exposures can bias diffusion coefficient estimates (up to 70% for particles with diffusion coefficients of 5 {micro}m2/s) and distort jump-distance distributions, underscoring the need for photon-by-photon tracking in fast-diffusion regimes. Moreover, SP2T delivers substantial computational gains--achieving more than a 50-fold GPU speedup over CPU-based likelihood tracking methods that assume continuous intensity, when compared on datasets with the same frame size and number of frames. Together, these advances establish SP2T as a robust, data-efficient solution for unbiased particle tracking with millisecond-to-microsecond temporal resolution.

3D tracking and localization of particles, typically fluorescently labeled biomolecules, provides a direct means of monitoring cellular transport and communication. However, sample-induced wavefront distortions of emitted fluorescent light as it passes through the sample and onto the detector often yield point spread function (PSF) aberrations, presenting an important challenge to 3D particle tracking using pre-calibrated PSFs. PSF calibration is typically performed outside cellular samples, ignoring sample-induced aberrations, which can result in localization errors on the order of tens to hundreds of nanometers, ultimately compromising sub-diffraction limited tracking. In practice, correcting sample-induced aberrations currently requires sample-specific hardware adjustments, such as adaptive optics. Yet, information on sample-induced aberrations and PSF shape can be directly decoded from data collected using a 3D imaging setup. To this end, we propose a framework for simultaneous particle tracking, pupil function learning, and PSF reconstruction directly from the input data themselves. To accomplish this, we operate within a Bayesian paradigm, placing continuous 2D priors on all possible pupil phase and amplitudes warranted by the data without limiting ourselves to a finite Zernike set-thereby allowing capture of intricate pupil phase details. We benchmark our framework using a wide range of synthetic and experimental data from static to diffusing particles, and generalize to multiple diffusing particles with overlapping PSFs. Further, as a result of simultaneous particle tracking, phase retrieval, and PSF reconstruction, we retrieve the pupil phase with errors smaller than 10% under a range of realistic scenarios, while restoring sub-diffraction limited localization precisions of 10-25 nm and 20-50 nm in lateral and axial directions, respectively.

With a limited dynamic range of an imaging system, there are always regions with signal intensities comparable to the noise level, if the signal intensity distribution is close to or even wider than the available dynamic range. Optical brain/neuronal imaging is such a case where weak-intensity ultrafine structures, such as, nerve fibers, dendrites and dendritic spines, often coexist with ultrabright structures, such as, somas. A high fluorescence-protein concentration makes the soma order-of-magnitude brighter than the adjacent ultrafine structures resulting in an ultra-wide dynamic range. A straightforward enhancement of the weak-intensity structures often leads to saturation of the brighter ones, and might further result in amplification of high-frequency background noises. An adaptive illumination strategy to real-time-compress the dynamic range demands a dedicated hardware to operate and owing to electronic limitations, might encounter a poor effective bandwidth especially when each digitized pixel is required to be illumination optimized. Furthermore, such a method is often not immune to noise-amplification while locally enhancing a weak-intensity structure. We report a dedicated-hardware-free method for rapid noise-suppressed wide-dynamic-range compression so as to enhance visibility of such weak-intensity structures in terms of both contrast-ratio and signal-to-noise ratio while minimizing saturation of the brightest ones. With large-FOV aliasing-free two-photon fluorescence neuronal imaging, we validate its effectiveness by retrieving weak-intensity ultrafine structures amidst a strong noisy background. With compute-unified-device-architecture (CUDA)-acceleration, a time-complexity of <3 ms for a 1000x1000-sized 16-bit data-set is secured, enabling a real-time applicability of the same.

Laser Speckle Contrast Imaging (LSCI) is a non-contact, label-free optical technique widely used in biomedical research and clinical applications. It enables real-time visualization and quantification of microvascular blood flow by analyzing the temporal fluctuations of laser speckles induced by moving red blood cells. However, conventional LSCI uses visible or near-infrared illumination, which--during prolonged exposure (e.g., >1{square}hr)--can induce sublethal neural stress and cause signal drift, compromising physiological relevance and raising ethical concerns. To mitigate these limitations, we introduce TunLSCI--a TransUNet-based recovery network designed to reconstruct high-fidelity mouse cerebral blood flow (CBF) indices from ultra-low-illumination LSCI. We train our network on paired ultra-low-illumination (1.27 {micro}W/mm2) and conventional LSCI data ([~]200 {micro}W/mm2 illumination, the latter as reference), and demonstrate that it outperforms the conventional standard analytical LSCI processing pipeline based on stLASCA, particularly in reconstructing fine vasculature from few frames, suppressing speckle noise, and maintaining robustness against exposure variations. We validate that the proposed TunLSCI reduces illumination power density by [~]157-fold compared with conventional stLASCA, well below the safety threshold for cortical exposure in mice and markedly improves stability during a 2-hour continuous mouse CBF monitoring. Our method significantly minimizes the phototoxic burden of LSCI while preserving spatiotemporal fidelity and quantitative accuracy, thus enabling longitudinal, high-biosafety cerebral perfusion tracking in vivo over multi-hours.

Imaging biological systems at subcellular resolution and across scales is essential to under-standing how cells form tissues, organs, and organisms. However, existing large-scale optical techniques often require harsh tissue-clearing methods that cause significant morphological changes, compromise the integrity of cell membranes, and reduce the signal of fluorescent proteins. Here, we demonstrate multifocal two-photon microscopy that enables imaging mesoscopic scattering samples in their native tissue environment at high resolution and high speed.

Optimal imaging strategies remain underdeveloped to maximize information for fluorescence microscopy while minimizing the harm to fragile living systems. Taking hint from the supercontinuum generation in ultrafast laser physics, we generated supercontinuum fluorescence from untreated unlabeled live samples before nonlinear photodamage onset. Our imaging achieved high-content cell phenotyping and tissue histology, identified bovine embryo polarization, quantified aging-related stress across cell types and species, demystified embryogenesis before and after implantation, sensed drug cytotoxicity in real-time, scanned brain area for targeted patching, optimized machine learning to track small moving organisms, induced two-photon phototropism of leaf chloroplasts under two-photon photosynthesis, unraveled microscopic origin of autumn colors, and interrogated intestinal microbiome. The results enable a facility-type microscope to freely explore vital molecular biology across life sciences.