Fluorescence microscopy has emerged as an indispensable tool in neuroscience research, enabling subcellular-resolution imaging of neural circuits and monitoring of brain function in vivo. However, physiological motion artifacts which include both rigid displacements and non-rigid deformations induced by respiration and tissue dynamics, significantly compromise imaging fidelity and quantitative analysis reliability. In this paper, we propose DeepMoCo, an innovative deep learning-based framework that integrates a graph neural network (GNN) to establish spatiotemporal feature correlations between biological motion patterns and imaging artifacts. This architecture enables rapid tracking and adaptive correction of complex motion patterns, achieving 95
Abstract Inspired by the key elements and principles in the brain, photonic neuromorphic computing shows great potential for building next-generation intelligent processing systems with high parallelism, low latency, low power consumption, and self-learning capabilities. While summarizing significant advances in this field, this review offers insights into how photonic neuromorphic computing systems support next-generation intelligent information processing. Specifically, we discuss emerging materials and devices, which support more compact integration of efficient physical architectures. Architectures such as photonic spiking neural networks and reservoir computing, together with associated learning paradigms, show a trend of collaborative development. Then we present promising applications, where neuromorphic computing systems provide broadband and multi-domain perception and processing. Finally, we discuss key challenges and future directions. This review aims to offer a clear and comprehensive overview for researchers across a broad community. We hope to present these insights in a “neuromorphic” manner, to reveal why learning from the brain has become increasingly important, especially for overcoming the efficiency bottleneck in traditional von Neumann architecture and data-intensive artificial neural networks.
Abstract Fluorescence lifetime imaging microscopy (FLIM) is a powerful quantitative technique that provides metabolic and molecular contrast, offering strong translational potential for label-free, real-time diagnostics. However, its clinical adoption remains limited by long pixel dwell times and low signal-to-noise ratio (SNR), which impose a more severe resolution-speed trade-off than conventional optical imaging approaches. Here, we introduce FLIM PSR_k , a deep learning-based multi-channel pixel super-resolution (PSR) framework that reconstructs high-resolution FLIM images from data acquired with up to a fivefold increased pixel size. The model is trained using the conditional generative adversarial network (cGAN) framework, which, compared to diffusion model-based alternatives, delivers a more robust PSR reconstruction with substantially shorter inference times, a crucial advantage for practical deployment. FLIM PSR_k not only enables faster image acquisition but can also alleviate SNR limitations in autofluorescence-based FLIM. Blind testing on held-out patient-derived tumor tissue samples demonstrates that FLIM PSR_k reliably achieves a super-resolution factor of k = 5, resulting in a 25-fold increase in the space-bandwidth product of the output images and revealing fine architectural features lost in lower-resolution inputs, with statistically significant improvements across various image quality metrics. By increasing FLIM’s effective spatial resolution, FLIM PSR_k advances lifetime imaging toward faster, higher-resolution, and hardware-flexible implementations compatible with low-numerical-aperture and miniaturized platforms, better positioning FLIM for translational applications.
Abstract Optical coherence is a fundamental yet underexploited degree of freedom for controlling light–matter interactions, with far-reaching implications for imaging, information processing, and photonic computing. Despite decades of progress, dynamic coherence control has remained constrained by an inherent trade-off: low efficiency, bulky system footprints, and limited dynamic tunability. These limitations severely constrain the integration and miniaturization of coherence-engineered photonic systems. Here, we propose and experimentally demonstrate a dynamic Pancharatnam–Berry phase nematic liquid–crystal device, photopatterned with programmable ultraviolet polarization, which addresses several key challenges in conventional coherence-control approaches, including compactness, reversibility, and dynamic tunability. By exploiting the electro-optical reorientation of soft-matter liquid crystals, our platform enables on-demand, reversible modulation of optical coherence within a millimeter-scale device, achieving a modulation efficiency up to 70% and a response time of 20 ms. This approach realizes continuous and fully dynamic coherence tuning, spanning the entire range from nearly coherent to nearly incoherent illumination. Beyond its fundamental significance, the demonstrated capability enables enhanced optical performance in scattering environments, including speckle-free imaging and dynamically reconfigurable “smart-window” functionality. Our results establish liquid–crystal–enabled coherence engineering as a compact, efficient, and scalable solution, opening a practical pathway toward integrated and miniaturized photonic systems for next-generation imaging, encryption, and photonic computing.
Abstract Computational spectral imaging overcomes the trade-off between spectral resolution and light throughput, but its performance still remains fundamentally limited by the geometric separability of spectral signatures in the encoding space. Current encoder designs predominantly focus on minimizing correlation or coherence, often overlooking the direct maximization of distinguishability among diverse spectra. Here we show that maximizing the minimum Euclidean distance in the encoding space (EDE) serves as a physically direct criterion for designing high-fidelity spectral encoders. To navigate the complex, high-dimensional design space under fabrication constraints, we developed a stochastic-deterministic optimization framework that couples global exploration via genetic algorithms with efficient local refinement using automatic differentiation on the transfer-matrix method. We demonstrate that EDE-optimized encoders significantly extend the minimum distance between distinct spectra in encoding space, yielding improvements in peak signal-to-noise ratio of 3.60 dB, 2.82 dB, and 4.52 dB compared with random, correlation-optimized, and coherence-optimized benchmarks, respectively, and achieving a reconstruction spectral fidelity exceeding 98.63% across diverse targets. Our approach enables high-precision spectral reconstruction, establishing a new geometric paradigm for the design of intelligent computational optical systems.
Abstract Recently, Prof. Wei Zheng and colleagues from Sun Yat-sen University have developed a novel nano-polycrystalline hexagonal boron nitride film material, featuring both large-area preparation capability and high luminescence efficiency. This is the first reported pure boron nitride neutron imaging material, which is capable of overcoming the limitations of traditional materials in spatial resolution and gamma-ray suppression. This advancement provides a pioneering impetus for the development of neutron imaging technology toward cutting-edge applications such as microscopic structural observation and radioactive substance detection.
Quantum secret sharing (QSS), as a fundamental cryptographic protocol for future quantum networks, continues to face significant challenges, particularly in the generation of multipartite entanglement and the degradation of entanglement fidelity during distribution, both of which severely limit its scalability. These persistent constraints motivate an alternative approach based on continuous-variable (CV) systems. We propose a CVQSS scheme based on an electro-optically modulated optical frequency comb. The scheme employs a single laser to generate multi-wavelength coherent states, enabling the efficient and flexible construction of secret sharing subnetworks. By incorporating a broadcast-based distribution mechanism, the architecture is scalable to 128 players. Experimental verification with 24 players over a 10 km fiber link demonstrates a secret sharing rate of 6.24 per player under asymptotic conditions and 1.27 Mbps per player under finite-size effects, and a 1.05MB image is secretly shared to six players. This work achieves information-theoretically secure QSS both within and across subnetworks, providing a solid technical foundation for the development of scalable and multifunctional quantum networks.
Sensing of heavy metals in coal ash is urgent for human health and environmental protection. However, conventional methods require chemical digestion, which cannot achieve both real-time and accuracy. Here, we demonstrate a novel optical sensing approach, called laser co-sourced plasma emission–absorption spectroscopic dual detection (LCSP-EASD), which innovatively couples the complementary and strongly correlated emission and absorption spectra of the same plasma. The emission spectrum mode enables the sensor to have comprehensive multi-element analytical capabilities, while the absorption spectrum mode enhances the detection sensitivity by utilizing the specific absorption of elements. The synergistic effect significantly enhances the performance of comprehensive, specific, real-time, and sensitive sensing of heavy metals. The experimental measured heavy metals in the pyrolysis products under different conditions closely matched the standard results, which verified the feasibility to detect heavy metals in coal ash. The results demonstrate that the limits of detection for Cr, Pb, Se, and Hg decreased to 2.14, 2.24 ppm, 464, and 119 ppb, respectively, which meet the standard limit of China (GB15618-2018). Compared with the TXRF, it has a 2.3 times enhancement in optimal sensitivity and 2 orders of magnitude shortened in experiment duration. Furthermore, comparisons with reference methods (< 5
Abstract Avalanche photodiodes (APDs) featuring internal multiplication hold great promise for high-sensitivity detection in optical systems. The gain-bandwidth product (GBP) comprehensively characterizes APDs’ overall performance, intrinsically linked to the effective impact ionization ratio, namely the k factor. An ultralow k factor suggests an ultrahigh GBP in theory, underscoring that one of the keys for the APD design should focus on minimizing the k factor. Here, we propose and experimentally demonstrate a novel waveguide Ge/Si APD with a record-high GBP of 5580 GHz (more than 500% higher than those reported previously), corresponding to a high gain of 209 and a bandwidth of 26.7 GHz when operating under −30 dBm input optical power. Introducing a PN junction as the multiplication region leads to a highly non-uniform electric field distribution for impact ionization and enables impact-ionization engineering without requiring complex multi-layer epitaxy processes. The effective k factor is measured to be as low as 0.011, which is even much lower than the intrinsic value of bulk silicon. With the present APDs, very high-speed PAM4 signals with a bit rate of up to 160 Gbps are received successfully with a clear eye-diagram. The present APD is further packaged with a transimpedance amplifier chip for 50G-PON systems, successfully achieving a high sensitivity of −26.5 dBm for 50-Gbps NRZ signals. Such high-performance Ge/Si waveguide APDs developed with simple fabrication processes show great potential for photodetection across diverse application scenarios.
Neural activity unfolds across three-dimensional circuits on millisecond-to-microsecond timescales, yet most optical microscopes still acquire volumes sequentially, limiting their ability to capture fast, distributed dynamics. Light-field microscopy (LFM) addresses this unmet need by encoding spatial and angular information into a single camera exposure, enabling snapshot volumetric imaging with low latency and strong robustness to motion. Here we review emerging advances in light-field neuroimaging, from brain-wide calcium recordings in freely moving animals to recent progress that brings kilohertz-class volumetric voltage imaging within reach. We argue that LFM should be evaluated based on information throughput, latency, photon efficiency, and motion robustness at the speed frontier, but not as a direct resolution or contrast competitor to confocal, multiphoton, or light-sheet microscopy. We conclude by highlighting future directions that preserve the LFM’s snapshot advantage, including speed-preserving improvements in image quality, extreme temporal-bandwidth architectures that prioritize quantitative inference over visual appearance, and multimodal light-field sensing that adds spectral, lifetime, and polarization contrast.
Abstract In-situ spectroscopy serves as a critical bridge between laboratory analysis of returned samples and orbital remote sensing of planetary surfaces. During China's Chang'e-6 (CE-6) mission, the first mission to return samples from the lunar farside, the Lunar Mineral Spectrometer (LMS) experienced internal temperatures exceeding 74 °C, threatening the fidelity of spectral measurements and undermining cross-scale comparisons. Here, we develop a Temperature-Compensated Radiometric Calibration (TCRC) framework to correct thermally induced measurement deviations. The framework improves the consistency of repeated observations acquired at different instrument temperatures by approximately 65%, while maintaining a signal-to-noise ratio above 40 dB under peak-temperature conditions. The corrected dataset enables cross-scale validation of surface properties at the lunar farside landing area. FeO abundance retrieved from LMS (15.36 – 19.54 wt%) agrees with both orbital Kaguya MI data (17.42 wt%) and returned sample analyses (~17.2 wt%). Centimeter-scale mapping on the Chang’e-6 landing area further reveals spatial heterogeneity in optical maturity and water content associated with lander plume disturbance. These results demonstrate that the TCRC framework supports reliable in-situ spectral acquisition under thermal extremes lunar surface conditions. Furthermore, the corrected spectra enable cross-scale validation and provide a practical reference for future planetary spectroscopic payloads operating under thermally challenging environments.
Integrated electro-optic (E-O) frequency combs built on the lithium niobate-on-insulator (LNOI) platform have emerged as a promising tool for diverse applications. Compared with high-quality-factor microresonator schemes, waveguide-based E-O combs deliver enhanced flexibility and higher efficiency. However, they often exhibit restricted spectral bandwidth due to their non-resonant optical characteristics. In this paper, we present a broadband waveguide-based E-O comb by using mode circulation. The inherent mode hybridization of the anisotropic LNOI waveguide is suppressed effectively by employing a novel Z-propagation designed mode multiplexer, thereby enabling the scaling of the mode-circulating E-O comb to four mode-channels (TE0, TE1, TE2, and TE3 modes). By integrating the mode-circulating scheme with a GSG traveling-wave electrode configuration, and carefully designing multimode phase modulators and delay line waveguides for each optical loop, the modulation index of the E-O comb is enhanced by a factor of eight. Experimentally, we successfully generated 128 comb lines covering a wavelength range of 25.6 nm by driving the fabricated E-O comb with a radio frequency (RF) signal of 25 GHz and power of 37 dBm. The comb exhibits a modulation index enhancement factor of approximately 7.5, an effective half-wave voltage of 1.24 V, and an optical loss of 7 dB. Spectral measurements are carried out by utilizing the reconfigurability of the present E-O comb, and a spectral resolution of 2 MHz ( 0.016 pm) is achieved within the wavelength range of 1530–1570 nm. Furthermore, the time-to-frequency mapping method is used to generate a broadband flat-top comb with 41 lines and 4 dB flatness. The present E-O comb offers the advantages of high energy efficiency, broad bandwidth, and reconfigurability, and will play a vital role in a wide range of applications, such as high-speed optical communications, high-precision measurements, and optical computing.
Abstract As AI data centers scale rapidly in throughput and size, the deployment and management complexity of fiber-connected interconnects is becoming a key bottleneck. Optical wireless communication (OWC) offers a flexible alternative to cabled interconnects, yet its per-channel data rates have historically lagged behind those of wired counterparts due to limited electro-optic bandwidth and front-end distortions. In this work, we narrow this gap through end-to-end co-design of the transceiver front ends, packaging interfaces, and nonlinear equalization algorithms, which together shape an effectively wideband, controlled-impedance electrical channel. Guided by systematic interface modeling, the proposed co-packaged solution eliminates the impedance discontinuity via dedicated transmission line design and an inductance-reduced bond-wire scheme, effectively extending the analog bandwidth while suppressing resonant peaking. To further mitigate residual level-dependent nonlinear intersymbol interference (ISI), an offline third-order Volterra-series nonlinear equalizer (VNE) is employed. Leveraging this hardware–algorithm synergy, we experimentally demonstrate a four-channel transceiver module that achieves record-high 100 Gb/s NRZ and 160 Gb/s PAM-4 per channel (640 Gb/s aggregate) over a 50 cm free-space link. To the best of our knowledge, this is the first multi-channel intensity-modulation and direct-detection (IM/DD) OWC system to attain wired-class performance, providing a scalable pathway for deploying high-speed OWC in next-generation data centers.
Abstract Conventional second-harmonic generation (SHG) microscopy fundamentally lacks the phase sensitivity, which is essential for characterizing symmetry properties that depend on phase information. Although interferometric SHG (I-SHG) enables phase retrieval through coherent superposition with an external reference beam, standard implementations based on phase-difference modulation face inherent limitations in microscopic systems due to optical aberrations, dispersion artifacts, and polarization constraints. Here, we introduce an angular-polarization-engineered I-SHG microscopy technique for phase-resolved imaging, offering minimal temporal walk-off, uniform phase distribution, and freedom from sample orientation and polarization restrictions. Instead of actively scanning the reference–sample phase difference with external phase-control elements, we tune the polarization-angle-dependent projection of the interference term by rotating the monolayer reference, thereby enabling stable and high-contrast phase-sensitive readout. Specifically, a thin two-dimensional material reference attached to a transparent substrate is placed between the objective and the sample in a sample-facing configuration. This setup enables the aforementioned advantages while maintaining diffraction-limited resolution and compatibility with cryo-vacuum conditions. Demonstrated through discrimination of inversion twin domains (intrinsic π-phase contrast), this approach achieves non-destructive phase-sensitive imaging at micron and sub-micron structural scales. The work establishes a versatile platform for quantitative symmetry metrology in nonlinear optical microscopy.
Thermal emission from natural materials is typically broadband and weakly structured in angle and polarization, limiting its utility for customized infrared functionalities. Here we demonstrate a metasurface platform that realizes broadband polarization-asymmetric directional thermal emission across the 8–14 μm long-wave infrared (LWIR) atmospheric window by engineering opposite reflection phase gradients for two arbitrary orthogonal polarization bases. The metasurface comprises anisotropic metallic meta-atoms integrated with a lossy multilayer film that supports low-Q gap-plasmon resonances, enabling a well-resolved polarization-dependent phase gradient throughout the LWIR band. Beyond a critical angle, the phase gradient supplies the in-plane momentum required to unidirectionally excite evanescent waves, leading to strongly polarization-dependent asymmetric absorption and thus asymmetric emission via Kirchhoff’s law. Leveraging a Particle Swarm Optimization (PSO) algorithm, we optimize supercells for both linear and circular orthogonal bases, and demonstrate that one polarization preferentially emits toward positive angles while the orthogonal polarization emits toward negative angles, achieving spatial separation of thermal emission by polarization. Angle-resolved spectral measurements and polarization-resolved thermal imaging validate the predicted broadband asymmetry and its reversal between orthogonal polarizations. Our results showcase that gradient metasurfaces provide a powerful platform for multi-degree-of-freedom control of thermal emission through evanescent-wave engineering, and open a practical route to broadband, polarization-encoded thermal emission for applications including polarization-resolved infrared imaging, dual-channel thermal communication, and adaptive thermal signature control.
Paraxial optical skyrmions have attracted significant attention due to their ability to propagate in free space. Through high harmonic generation (HHG) and free-electron laser setups, spatial skyrmions with a topological plane perpendicular to the optical axis have been extended to the extreme ultraviolet (EUV) region, both theoretically and experimentally. More recently, spatiotemporal skyrmions with a topological plane parallel to the optical axis have been realized experimentally, though so far only in the visible and near-infrared regimes. EUV spatiotemporal skyrmions hold promise due to higher photon energy and potential as a novel attosecond light source. In this work, for the first time, we theoretically demonstrate the generation and control of spatiotemporal skyrmions in the EUV region via HHG. Harmonic fields driven by two-color lasers exhibit Bloch-type skyrmionic characteristics, and varying laser parameters allow other EUV skyrmions and vector hopfions to be generated. A photon absorption model explains the coherent transfer of spatiotemporal characteristics from driving fields to harmonics, with Stokes-vector direction matching being crucial. Furthermore, we have proposed a spatiotemporal mode conversion (STMC) scheme that can recover the near-field spatiotemporal structure of the spatiotemporal Skyrmions in the far field. This research expands the family of EUV structured optical fields and lays the groundwork for experimental generation and far-field applications of EUV spatiotemporal skyrmions and vector hopfions.
Abstract Hyperspectral imaging has long been constrained by inherent trade-offs among spatial, spectral, and temporal resolutions, as well as by limited optical throughput and reliance on complex micro-nano fabrication processes. Here, we introduce a paradigm-shifting hyperspectral imaging system that fundamentally overturns these constraints by decoupling spectral modulation from detection. Central to this system is a spectrally programmable organic light-emitting diode (OLED) that enables continuous spectral tuning across the entire visible spectrum within a single, monolithic pixel. This breakthrough is achieved via a novel charge-generation layer architecture that integrates two vertically stacked, oppositely oriented p–i–n units, allowing for precise selection and superposition of emissive channels under an alternating current field. This OLED serves as an ideal illumination source for hyperspectral imaging, capable of emitting tunable narrowband light or complex, superimposed spectral profiles. When integrated with a standard CMOS sensor and reconstruction algorithms, this decoupled architecture achieves hyperspectral imaging with a 2 nm spectral resolution dictated by the OLED, along with a 3072 × 2048 spatial resolution and 60.9 fps raw full-resolution frame rate, both dictated by the CMOS sensor, all within a compact footprint. This work establishes a transformative path for high-performance hyperspectral imaging, promising broad applicability in machine vision, consumer electronics, and beyond.
Abstract Graphene has led the exploration of nonlinear optical responses in two-dimensional materials with exceptionally strong third-order nonlinearity and its electrical controllability. Nonlinear wave mixing with difference-frequency is particularly interesting in graphene because of the divergent nature of third-order susceptibility as the frequency difference approaches zero, but the study on nearly degenerated four-wave mixing (NDFWM) process in graphene is largely unexplored. In this work, we report the giant third-order susceptibility of monolayer graphene, reaching the order of 10–13 m2 V−2 at the optical telecom C-band via the NDFWM process, and its electrical tunability with a high on–off contrast of 23 dB. Moreover, we observed that the NDFWM response under electrical doping exhibits a resonance feature at low pump intensity in ambient conditions, which is substantially altered by varying the pump power. Through non-perturbative quantum master equation calculations, we revealed that our observation is closely related to the dephasing nature of the Dirac fermion of graphene. The decoherence time of photoexcited carriers is estimated up to 70 fs at low pump intensity, which regime is not accessible by other nonlinear means such as high harmonic generation requiring high intensity light. Our findings not only pave an unprecedented route for probing nonlinear dynamics of photoexcited carriers across a wide range but also have a significant impact on ultrafast nonlinear information processing in graphene.
Coupled resonator optical waveguides (CROWs) are widely used for filtering, storing, and enhancing light-matter interactions. However, their bandwidth, which characterizes the operation wavelength range, is fundamentally limited by the tradeoff between the free spectral range (FSR) and finesse. To address this limitation, we introduce topological coupler into CROWs, significantly enlarging their bandwidth. By leveraging the topology and symmetry of topological coupler, the finesse of CROW is independent of its FSR, which efficiently relieves the tradeoff. Our topological CROW achieves a bandwidth of 9.0 nm. By cascading such CROWs, we realize a topological two-channel add-drop filter with a bandwidth of over 4.7 nm, even with dimensional errors up to 20 nm. Furthermore, high-speed data transmission at 170 Gb/s is realized in the topological filter. Our broadband topological CROWs and filter offer new opportunities for robust optical buffering, broadband modulation, and novel polariton applications.
Abstract The quantitation detection of breast and ovarian cancer markers hold crucial significance in early diagnosis and screen for women’s health. However, there still lacks of a rapid and highly sensitive strategy for multiplexed detection of female-related tumor markers. Herein, we present a strategy for tuning and selectively enhancing upconversion (UC) emission colors by integrating a single core–shell upconversion nanoparticle (UCNP) with a photonic crystal (PC) cavity. Experimental and theoretical findings reveal that the excitation optical filed modulation surrounding single UCNP dominates the variation of UC emission through PC effect. Subsequently, we designed a PC cavity enhanced single UCNP microfluidic device for binary-channel tumor markers based on fluorescence resonance energy transfer effect. As a proof of concept, this device was employed for the quantitative detection of breast cancer biomarker miRNA-155 and the multiplexed detection of ovarian cancer biomarkers with high sensitivity and specificity. The biosensor achieved wide linear detection ranges down to following concentrations: miRNA-155 (1 pM-0.6 nM), CA125 (0.002–1.8 U/mL), and HE4 (0.008–85.5 ng/mL). This work offers a powerful strategy for manipulating UC of single particle and opens new opportunities for highly sensitive, multi-channel detection of various tumor markers.
