3 February 2025 in Applied Physics Letters
We demonstrate a 100 mm-diameter, 2.4 μm-thick multilevel diffractive lens (MDL) with a 200 mm focal length, optimized for the 400 to 800 nm wavelength range—specifications that are difficult to achieve even with complex multi-element refractive systems. Created using an inverse-design approach and grayscale lithography, the MDL achieves achromatic focusing, confirmed through hyperspectral point-spread function (PSF) characterization. Imaging experiments resolved spatial frequencies up to 181 lp/mm and demonstrated the MDL's capability in capturing high-quality, full-color images of the moon, sun, and distant terrestrial scenes. Color-enhanced lunar images revealed key geological features, while solar imaging identified visible sunspots. Additionally, the MDL was integrated with a refractive achromatic lens to form a hybrid telescope, significantly reducing weight for airborne and space-based imaging applications. Simulations and experimental results reported here underscore the potential of large-area achromatic flat lenses as lightweight alternatives to conventional refractive systems for astrophotography and other long-range imaging tasks.
1 February 2025 in Review of Scientific Instruments
Low-cost, minimally invasive microscopy for tracking cellular dynamics in living plants within their natural ecosystems is crucial for addressing fundamental questions in plant ecology and biology. However, existing solutions are constrained by coarse resolution, limited field-of-view (FoV), and poor deployability in natural settings. Here, we utilize a compact, portable microscope (“miniscope”) for label-free (autofluorescence) imaging in living poplar wood. We systematically implement and evaluate multiple computational methods to enhance resolution and FoV. Our optimal computational pipeline, comprising maximal intensity projection, deconvolution, and flat-field correction, increases resolution by up to 39% on-axis and up to 49% at the field edges, resolving features of 2.87 μm, averaged over a FoV of ∼1 mm (diameter), compared with a 4.34 μm baseline. We demonstrate microscopy within the tissue of a living poplar plant in our greenhouse, observing the embolism of vessel elements, wound response, and tissue deformation from moisture evaporation.
19 September 2024 in eLife
In the mammalian neocortex, inhibition is important for dynamically balancing excitation and shaping the response properties of cells and circuits. The various computational functions of inhibition are thought to be mediated by different inhibitory neuron types, of which a large diversity exists in several species. Current understanding of the function and connectivity of distinct inhibitory neuron types has mainly derived from studies in transgenic mice. However, it is unknown whether knowledge gained from mouse studies applies to the non-human primate, the model system closest to humans. The lack of viral tools to selectively access inhibitory neuron types has been a major impediment to studying their function in the primate. Here, we have thoroughly validated and characterized several recently developed viral vectors designed to restrict transgene expression to GABAergic cells or their parvalbumin (PV) subtype, and identified two types that show high specificity and efficiency in marmoset V1. We show that in marmoset V1, AAV-h56D induces transgene expression in GABAergic cells with up to 91–94% specificity and 79% efficiency, but this depends on viral serotype and cortical layer. AAV-PHP.eB-S5E2 induces transgene expression in PV cells across all cortical layers with up to 98% specificity and 86–90% efficiency, depending on layer. Thus, these viral vectors are promising tools for studying GABA and PV cell function and connectivity in the primate cortex.
26 August 2026 in PNAS Nexus
Living tissues contain dynamic biochemical information that is difficult to capture with conventional hyperspectral microscopes because sequential spectral acquisition is poorly matched to in vivo molecular processes that evolve during measurement. Here, we introduce a task-specific optical encoding framework for video-rate molecular inference in living plant tissue. The system integrates a passive spectral encoder, implemented here as a low-angle scattering low-density polyethylene (LDPE) layer, into a 22-mm miniaturized probe and learns a supervised mapping from ultraviolet-excited autofluorescence measurements to biomolecular abundance maps. Unlike conventional pipelines that first reconstruct hyperspectral datacubes and then perform spectral unmixing, the deployed system directly estimates endogenous molecular contrast associated primarily with lignin and chlorophyll in poplar tissue, with additional suberin-associated contrast evaluated in suberin-rich tissue. This reframing makes the measurement task biomolecular inference rather than spectral reconstruction, enabling biochemical mapping under low-photon autofluorescence conditions while reducing data burden and computational latency. In living poplar stems, the platform captures autofluorescence-derived videos of embolism propagation and wound-induced biochemical remodeling, dynamic processes for which sequential spectral acquisition can introduce temporal mixing because the molecular contrast evolves during the scan itself. The system also resolves genotype-dependent reductions in lignin-associated autofluorescence in engineered poplar lines. Direct molecular inference improves biomolecular estimation relative to a reconstruction-based pipeline, while probabilistic decoding provides uncertainty estimates. These results show that compact passive spectral encoding, when optimized for biological inference rather than datacube recovery, enables deployable, label-free molecular videography of living plant tissue dynamics after task-specific calibration.
9 July 2024 in Applied Optics
This study presents significant advancements in computational cannula microscopy for live imaging of cellular dynamics in poplar wood tissues. Leveraging machine-learning models such as pix2pix for image reconstruction, we achieved high-resolution imaging with a field of view of using a -core diameter probe. Our method allows for real-time image reconstruction at 0.29 s per frame with a mean absolute error of 0.07. We successfully captured cellular-level dynamics in vivo, demonstrating morphological changes at resolutions as small as . We implemented two types of probabilistic neural network models to quantify confidence levels in the reconstructed images. This approach facilitates context-aware, human-in-the-loop analysis, which is crucial for in vivo imaging where ground-truth data is unavailable. Using this approach we demonstrated deep in
vivo computational imaging of living plant tissue with high confidence (disagreement score ). This work addresses the challenges of imaging live plant tissues, offering a practical and minimally invasive tool for plant biologists.
26 August 2026 in Optics Express
A deep defogging pipeline pretrained on controlled laboratory fog and fine-tuned with domain-randomized synthetic fog applied to clear outdoor scenes generalizes across a graded sequence of out-of-distribution settings with no target-domain training, from chamber-free free-flowing fog to iPhone video recorded through an aircraft cabin window in flight, an entirely unseen sensor, scene, and optical path. This directly addresses an open transfer limitation reported for real-world defogging. Two design choices support the transfer. First, a single-camera fog imager photographs a flat-panel display through an artificial-fog enclosure with a fixed 114 mm scattering path, producing 5,495 pixel-aligned foggy/clear pairs. Exact registration permits a paired Laplacian ratio that predicts per-image restoration quality far better than single-image proxies (Spearman ρ = 0.632 versus 0.399) and supports pixel-exact L 1 reconstruction training that avoids adversarial hallucination. Second, the fog-chamber checkpoint is briefly fine-tuned on Mapillary Vistas crops overlaid with on-the-fly randomized synthetic fog spanning a broad range of strengths, spatial variations, airlights, and noise conditions to suppress oversaturation and tile-stitching artifacts during transfer. On a 552-image held-out split, a uniform comparison of 30 restoration backbones places NAFNet at the top (24.33 dB / 0.7912 SSIM), with a compact alternative within 1.29 dB at 3% of the parameter count, and a ResNet-50 classifier confirms that the restoration preserves semantic content rather than only pixel-level structure. On unpaired aircraft-window video, NIQE decreases from a mean of 6.22 to 4.97 after fine-tuning, with temporally stable output across full-motion sequences. The same backbone, under paired supervision, also reaches 20.71 dB / 0.683 SSIM on a non-overlapping O-HAZE/NH-HAZE split (a transferability check rather than a competitive ranking).