Journals >Advanced Photonics Nexus
Contents
2025
Volume: 4 Issue 6
16 Article(s)
Export citation format
Research Articles
Fluorescence microscopy image denoising via a wavelet-enhanced transformer based on DnCNN network
Shuhao Shen, Mingxuan Cao, Weikai Tan, E Du, and Xueli Chen
Fluorescence microscopy is indispensable in life science research, yet denoising remains challenging due to varied biological samples and imaging conditions. We introduce a wavelet-enhanced transformer based on DnCNN that fuses wavelet preprocessing with a dual-branch transformer–convolutional neural network (CNN) archFluorescence microscopy is indispensable in life science research, yet denoising remains challenging due to varied biological samples and imaging conditions. We introduce a wavelet-enhanced transformer based on DnCNN that fuses wavelet preprocessing with a dual-branch transformer–convolutional neural network (CNN) architecture. Wavelet decomposition separates high- and low-frequency components for targeted noise reduction; the CNN branch restores local details, whereas the transformer branch captures global context; and an adaptive loss balances quantitative fidelity with perceptual quality. On the fluorescence microscopy denoising benchmark, our method surpasses leading CNN- and transformer-based approaches, improving peak signal-to-noise ratio by 2.34% and 0.88% and structural similarity index measure by 0.53% and 1.07%, respectively. This framework offers enhanced generalization and practical gains for fluorescence image denoising..
Advanced Photonics Nexus
- Publication Date: Oct. 15, 2025
- Vol. 4, Issue 6, 066001 (2025)
Microscopic structured light 3D imaging via a scattering lens
Wenjing Zhao, Wei Chang, Youtao Wang, Aiping Zhai, Fei Liu, and Dong Wang
Transforming a scattering medium into a lens for imaging very simple binary objects is possible; however, it remains challenging to image complex grayscale objects, let alone measure 3D continuous distribution objects. Here, we propose and demonstrate the use of a ground glass diffuser as a scattering lens for imaging Transforming a scattering medium into a lens for imaging very simple binary objects is possible; however, it remains challenging to image complex grayscale objects, let alone measure 3D continuous distribution objects. Here, we propose and demonstrate the use of a ground glass diffuser as a scattering lens for imaging complex grayscale fringes, and we employ it to achieve microscopic structured light 3D imaging (MSL3DI). The ubiquitous property of the speckle patterns permits the exploitation of the scattering medium as an ultra-thin scattering lens with a variable focal length and a flexible working distance for microscale object measurement. The method provides a light, flexible, and cost-effective imaging device as an alternative to microscope objectives or telecentric lenses in conventional MSL3DI systems. We experimentally demonstrate that employing a scattering lens allows us to achieve relatively good phase information and robust 3D imaging from depth measurements, yielding measurement accuracy only marginally lower than that of a telecentric lens, typically within approximately 10 μm. Furthermore, the scattering lens demonstrates robust performance even when the imaging distance exceeds the typical working distance of a telecentric lens. The proposed method facilitates the application of scattering imaging techniques, providing a more flexible solution for MSL3DI..
Advanced Photonics Nexus
- Publication Date: Oct. 15, 2025
- Vol. 4, Issue 6, 066002 (2025)
Dual-split-ring resonant metasurface with both quasi-BIC and dipole modes for terahertz trace sensing of hyaluronic acid
Jiyue Chen, Jining Li, Kai Chen, Xiang Yang, Degang Xu, and Jianquan Yao
The resonance generated by different mechanisms naturally has different characteristics in sensing, and these differences increase the potential for specific detection. We designed a metasurface with both a quasi-bound state in continuum (QBIC) resonance and dipole resonance by conducting physical analyses such as elecThe resonance generated by different mechanisms naturally has different characteristics in sensing, and these differences increase the potential for specific detection. We designed a metasurface with both a quasi-bound state in continuum (QBIC) resonance and dipole resonance by conducting physical analyses such as electric field, current distribution, and multiple expansions on a dual-split-ring resonance with dipole resonance and a variant structure with symmetry breaking. On the other hand, the edge length of the slit was extended through a tilted split design, which further enhanced the QBIC resonance signal of the metasurface. In the sensing experiment of hyaluronic acid (HA), the limit of detection (LOD) obtained through frequency shift was 0.958 pmol / μL, whereas the LOD obtained through the change in transmittance was 0.02 pmol / μL. Our research findings contribute to the design of multiple resonant metasurfaces with different resonance modes, promoting further development in metasurface research and biosensing..
Advanced Photonics Nexus
- Publication Date: Oct. 16, 2025
- Vol. 4, Issue 6, 066003 (2025)
Customizing the field of view for imaging through scattering media
Dajiang Lu, Juncheng Chen, Yibin Tian, Jiapeng Cai, Zaoxin Chen, Xiang Peng, and Wenqi He
Noninvasive speckle autocorrelation is a promising technique for single-shot optical imaging through scattering media. However, it fails to image multiple distinct targets within an object space through scattering media because it is constrained by the tiny effective range of the optical memory effect. We present a metNoninvasive speckle autocorrelation is a promising technique for single-shot optical imaging through scattering media. However, it fails to image multiple distinct targets within an object space through scattering media because it is constrained by the tiny effective range of the optical memory effect. We present a method for multi-object single-shot imaging through scattering media that incorporates deep learning into the speckle autocorrelation technique, wherein the field of view (FOV) is customized by recovered autocorrelation sidelobes, and a conventional phase-retrieval algorithm is applied to a complete set of expected speckle autocorrelations to identify multiple target objects and their relative positions. Experiments verify the feasibility of customizing the FOV for imaging through scattering media. Image reconstruction results show that the proposed approach produces superior image quality compared to existing methods. We also demonstrate its generalization capability across different object types and unknown scattering media..
Advanced Photonics Nexus
- Publication Date: Oct. 22, 2025
- Vol. 4, Issue 6, 066004 (2025)
Index-adapting cladding light stripper for high-power thulium fiber lasers
Tilman Lühder, Till Walbaum and Thomas Schreiber
Cladding light strippers (CLSs) are essential components for high-power monolithic fiber laser systems. Because they allow for bending of the fiber, which leads to an excellent stripping efficiency of light with a low ray angle, refractive index-based CLSs have an advantage over the commonly used alternative approachesCladding light strippers (CLSs) are essential components for high-power monolithic fiber laser systems. Because they allow for bending of the fiber, which leads to an excellent stripping efficiency of light with a low ray angle, refractive index-based CLSs have an advantage over the commonly used alternative approaches. However, conventional high-index CLSs overheat at relatively low input power as the maximum temperature, located in a hot-spot, increases linearly with the input power. This applies particularly to CLSs in thulium-based fiber systems, where very low power can already lead to extreme heat generation due to the high cladding material absorption around 2 μm. Here, we investigate materials with a highly negative thermo-optical coefficient combined with a refractive index closely above glass to distribute the stripped power and heat uniformly along an increased fiber length. Analyzing multiple CLS geometries for fiber diameters of 125 and 400 μm, we show record-high maximum input powers for single-material CLSs of 21.8 W for the signal (2039 nm) and 675 W for the pump wavelength (793 nm). Transmitting excess light instead of overheating, this wavelength-adaptable self-protecting CLS concept is fast to apply onsite in the lab and reaches stripping efficiencies of >40 dB in the bent version..
Advanced Photonics Nexus
- Publication Date: Oct. 31, 2025
- Vol. 4, Issue 6, 066005 (2025)
Sub-diffraction-limited focusing laser pulses at ultrahigh intensity via near-critical-density hollow plasma fibers
Tianqi Xu, Zhuo Pan, Ying Gao, Yulan Liang, Shirui Xu, Qingfan Wu, Tan Song, Yujia Zhang, Haoran Chen, Qihang Han, Chenghao Hua, Ziyang Peng, Ke Chen, Xuan Liu, and Wenjun Ma
Ultra-intense electromagnetic fields exceeding 1023 W / cm2 are enabling breakthroughs in compact laser-driven particle accelerators and revealing new quantum electrodynamics (QED) phenomena. However, conventional laser-focusing methods face considerable engineering challenges and require substantial costs. Focusing scUltra-intense electromagnetic fields exceeding 1023 W / cm2 are enabling breakthroughs in compact laser-driven particle accelerators and revealing new quantum electrodynamics (QED) phenomena. However, conventional laser-focusing methods face considerable engineering challenges and require substantial costs. Focusing schemes utilizing plasma optics can produce sub-micrometer focus spots beyond the diffraction limit and substantially enhance the peak intensity; however, owing to significant energy dissipation, they may fail to simultaneously increase the laser fluence. To address these challenges, we propose a focusing scheme employing a near-critical-density hollow plasma fiber (HPF) that utilizes graded refractive index dynamics to boost both laser peak intensity and fluence at the same time. Three-dimensional particle-in-cell simulations demonstrate the HPF’s capability to focus a 4.5-μm-diameter Gaussian beam to a sub-diffraction-limited 0.6-μm-diameter spot. The peak intensity and laser fluence can be enhanced by factors of 22 and 10, respectively, marking a substantial improvement over existing plasma-based focusing schemes. Furthermore, the proposed scheme exhibits wide-range parameter adaptation and high robustness, making it suitable for direct implementation in PW-class ultra-intense laser experiments..
Advanced Photonics Nexus
- Publication Date: Nov. 07, 2025
- Vol. 4, Issue 6, 066007 (2025)
Research on the application of MobileNetV1 neural network model for small-sample OAM mode recognition
Yanyu Lu, Dahai Yang, Xikun Chen, Zhihao Xu, Wu Zhang, and Xianyou Wang
Deep learning (DL) models have demonstrated significant value in computational perception, super-resolution imaging, ultra-precision measurement, and photonic device design. In optical communication signal recognition, DL models can achieve fast and accurate identification. However, in high-capacity optical communicatiDeep learning (DL) models have demonstrated significant value in computational perception, super-resolution imaging, ultra-precision measurement, and photonic device design. In optical communication signal recognition, DL models can achieve fast and accurate identification. However, in high-capacity optical communication systems represented by orbital angular momentum (OAM) beams, neural networks often suffer from excessive parameter sizes and demand large training datasets. To address these challenges, we report a lightweight MobileNetV1 model optimized with efficient channel attention to perform OAM mode recognition after transmission through free space and underwater tank environments. Experimental results show that in simulated small-sample classification tasks with five samples per class, the proposed model achieves an accuracy of 99.67% even under moderate turbulence conditions, outperforming four other DL models. In addition, for experimental datasets collected from both atmospheric turbulence and underwater environments, the model consistently achieves recognition accuracies exceeding 90%, demonstrating strong generalization ability and a 77% reduction in parameter count compared to traditional convolutional neural network (CNN)-based DL models. We provide a new approach for deploying lightweight DL algorithms on resource-constrained embedded optical signal detection devices..
Advanced Photonics Nexus
- Publication Date: Nov. 08, 2025
- Vol. 4, Issue 6, 066008 (2025)
Spatially confined tumor phototherapy enabled by GHz laser thermal accumulation dynamics
Zhi Chen, Changle Meng, Huiling Lin, Dror Fixler, and Han Zhang
Achieving precise tumor ablation without damaging surrounding healthy tissue remains a significant challenge in cancer therapy, particularly for deep-seated or irregularly shaped tumors. Traditional laser-based approaches, although minimally invasive, are often limited by insufficient tissue penetration, uncontrolled tAchieving precise tumor ablation without damaging surrounding healthy tissue remains a significant challenge in cancer therapy, particularly for deep-seated or irregularly shaped tumors. Traditional laser-based approaches, although minimally invasive, are often limited by insufficient tissue penetration, uncontrolled thermal damage, and narrow therapeutic windows. We introduce GHz high-repetition-rate pulsed lasers as a transformative modality for tumor ablation. This approach capitalizes on the thermal accumulation effect of GHz pulse trains, in which the pulse interval is significantly shorter than the thermal relaxation time of biological tissue. Such a regime enables efficient and localized heat deposition in tumor regions. By precisely tuning the repetition frequency, pulse duration, and energy density, we establish a dynamic “ablation-cooling” cycle: rapid energy delivery followed by transient inter-pulse cooling. This thermal modulation ensures sharply confined ablation zones with reduced collateral damage. Our systematic investigation of laser-tissue interaction parameters demonstrates that GHz lasers offer superior spatial selectivity, minimized off-target injury, and enhanced treatment safety, presenting a compelling rationale for clinical translation of this paradigm in precision photothermal oncology..
Advanced Photonics Nexus
- Publication Date: Nov. 12, 2025
- Vol. 4, Issue 6, 066009 (2025)
Demonstration of the MICRO solar magnetograph using silicon nitride photonics and interferometric imaging
Humphry Chen, Shelbe Timothy, Neal Hurlburt, Gopal Vasudevan, Lawrence Shing, Tony Kowalczyk, and Simon Avery
We demonstrate a silicon nitride photonics-based imaging system that can perform one-dimensional interferometric imaging around the 1550-nm wavelength. The magnetograph using interferometric and computational imaging for remote observations (MICRO) design uses silicon nitride on a Si platform to replace the bulky free-We demonstrate a silicon nitride photonics-based imaging system that can perform one-dimensional interferometric imaging around the 1550-nm wavelength. The magnetograph using interferometric and computational imaging for remote observations (MICRO) design uses silicon nitride on a Si platform to replace the bulky free-space optics of traditional magnetograph imaging systems with nanofabricated structures of a fraction of the size. The photonic integrated circuit (PIC) uses an array of lenslets that couple light into four input waveguides with spacing arranged along a Golomb ruler, where each aperture pair formed has a unique length. Each aperture is mixed with a 13-dBm reference laser and separated inside a 2 × 4 optical hybrid to generate in-phase and quadrature-phase signals to be detected in balanced detectors at the output of the PIC. We use a field programmable gate array (FPGA) board to digitize and process the measurements. The FPGAs and PIC are combined to reduce the overall size, weight, and power of the system, paving the way for a compact imaging system. We demonstrate a PIC-based imager design and experimental testbed for spectrometry applications..
Advanced Photonics Nexus
- Publication Date: Nov. 19, 2025
- Vol. 4, Issue 6, 066010 (2025)
Comprehensive experimental evaluation of temporal contrast enhancement techniques applied to the petawatt-class J-KAREN-P laser system
Hiromitsu Kiriyama, Akito Sagisaka, Yasuhiro Miyasaka, Akira Kon, Mamiko Nishiuchi, Alexander S. Pirozhkov, Yuji Fukuda, Koichi Ogura, Kotaro Kondo, Nobuhiko Nakanii, Yuji Mashiba, Nicholas P. Dover, Liu Chang, Stefan Bock, Tim Ziegler, Thomas Püschel, Karl Zeil, Ulrich Schramm, Il Woo Choi, and Chang Hee Nam
We present a systematic experimental investigation of temporal contrast enhancement techniques for petawatt (PW)-class Ti:sapphire lasers utilizing a double chirped-pulse amplification (CPA) architecture. Particular attention is given to pre-pulses induced by post-pulses originating in the first CPA stage. One conventiWe present a systematic experimental investigation of temporal contrast enhancement techniques for petawatt (PW)-class Ti:sapphire lasers utilizing a double chirped-pulse amplification (CPA) architecture. Particular attention is given to pre-pulses induced by post-pulses originating in the first CPA stage. One conventional and two advanced pulse-cleaning strategies are quantitatively evaluated: (i) a saturable absorber (SA), (ii) a femtosecond optical parametric amplifier (OPA) employing the idler pulse in a two-stage configuration, and (iii) sum-frequency generation (SFG) combining the signal and idler pulses from the OPA. All techniques are implemented and evaluated using the J-KAREN-P laser system with an output energy of about 20 J. To the best of our knowledge, this is the first report to directly and systematically compare the contrast of pre-pulses originating from the first CPA stage under identical experimental conditions in a high-energy PW-class laser facility. The results offer crucial insights into contrast optimization for future high-field applications..
Advanced Photonics Nexus
- Publication Date: Nov. 19, 2025
- Vol. 4, Issue 6, 066011 (2025)
High-fidelity and compact topology architecture for large-scale reconfigurable linear optical networks | Editors' Pick
Shuai Lin, Jinjie Zeng, Shuqing Lin, Siyuan Yu, and Yanfeng Zhang
Reconfigurable linear optical networks based on Mach-Zehnder interferometer (MZI) offer significant potential in optical information processing, particularly in emerging photonic quantum computing systems. However, device losses and calibration errors accumulate as network complexity grows, posing challenges in performReconfigurable linear optical networks based on Mach-Zehnder interferometer (MZI) offer significant potential in optical information processing, particularly in emerging photonic quantum computing systems. However, device losses and calibration errors accumulate as network complexity grows, posing challenges in performing precise mapping of matrix operations. Existing architectures, such as Diamond and Bokun, introduce MZI redundancy into Reck and Clements architectures to improve reliability, which increases complexity and differential path losses that limit scalability. We propose a compact topology architecture that achieves 100% fidelity by employing a symmetrical MZI to decouple optical loss from power ratio and introducing extra MZIs to enforce uniform loss distributions. This multi-level optimization enables direct monitoring pathways while supporting precise calibration, and it approaches theoretical fidelity in practical deployments with direct implications for scalable and fault-tolerant photonic computing systems..
Advanced Photonics Nexus
- Publication Date: Nov. 19, 2025
- Vol. 4, Issue 6, 066012 (2025)
Compressive incoherent digital holography for high-fidelity 3D imaging
Ning Xu, Dalong Qi, Long Cheng, Zhen Pan, Wenzhang Lin, Chengyu Zhou, Hongmei Ma, Yunhua Yao, Yuecheng Shen, Lianzhong Deng, Zhenrong Sun, and Shian Zhang
Incoherent digital holography has attracted significant attention due to its advantages in three-dimensional (3D) imaging under low spatial coherence conditions, such as easy access to light sources and reduced speckle noise. However, interlayer crosstalk during the reconstruction process leads to a substantial reductiIncoherent digital holography has attracted significant attention due to its advantages in three-dimensional (3D) imaging under low spatial coherence conditions, such as easy access to light sources and reduced speckle noise. However, interlayer crosstalk during the reconstruction process leads to a substantial reduction in reconstruction fidelity. Furthermore, existing deconvolution- and deep-learning-based reconstruction algorithms face limitations in terms of effectiveness and generalization. To address these challenges, we propose a compressive incoherent digital holography (CIDH) approach for 3D imaging. In CIDH, a point spread hologram sequence with a high signal-to-noise ratio is initially obtained using a customized computer-generated holography method for dual-channel forward data acquisition. For scene reconstruction, a compressed sensing-based two-step iterative shrinkage/thresholding algorithm is employed to achieve high-fidelity 3D scene retrieval. The combined optimization demonstrates exceptional performance in suppressing interlayer crosstalk and enhancing reconstruction fidelity. In simulations, crosstalk was effectively suppressed across 10 depth layers. In experiments, successful suppression was achieved for both a five-layer transmissive object and a two-layer reflective 3D object, resulting in significantly improved reconstruction accuracy. The proposed framework shows great potential for applications in various incoherent source-illuminated and fluorescent 3D imaging..
Advanced Photonics Nexus
- Publication Date: Nov. 20, 2025
- Vol. 4, Issue 6, 066013 (2025)
Implicit neural representation based on optoelectronic periodic nonlinear activation
Jiawei Gu, Yulong Huang, Zijie Chen, Mu Ku Chen, and Zihan Geng
Implicit neural representation (INR) networks break through the accuracy and resolution limitations of traditional discrete representations by modeling high-dimensional data as continuously differentiable implicit neural networks, enabling lossless compression and efficient reconstruction of details in a compact form. Implicit neural representation (INR) networks break through the accuracy and resolution limitations of traditional discrete representations by modeling high-dimensional data as continuously differentiable implicit neural networks, enabling lossless compression and efficient reconstruction of details in a compact form. However, an optical-assisted INR network has yet to be demonstrated. INR networks require high nonlinearity, whereas implementing analog nonlinear activation in photonic neural networks is a challenge. Inspired by the inherent physical properties of modulators, we propose an optoelectronic nonlinear activation and implement it on the image reconstruction task. Simulations and experiments demonstrate that the proposed optoelectronic periodic neural network can represent images and perform image reconstruction with excellent results. This approach empowers complex image reconstruction with high-frequency details and reduces the amount of required hardware. Our method enables the development of compact, efficient optoelectronic neural networks, utilizing repeatable modular units for scalable and practical high-performance computing. It can enable scene generation and compression in biomedicine, autonomous driving, and augmented reality/virtual reality..
Advanced Photonics Nexus
- Publication Date: Nov. 20, 2025
- Vol. 4, Issue 6, 066014 (2025)
Elliptical vectorial metrics for physically plausible polarization information analysis
Runchen Zhang, Xuke Qiu, Yifei Ma, Zimo Zhao, An Aloysius Wang, Jinge Guo, Ji Qin, Steve J. Elston, Stephen M. Morris, and Chao He
The Mueller matrix polar decomposition method decomposes a Mueller matrix into a diattenuator, a retarder, and a depolarizer. Among these elements, the retarder, which plays a key role in medical and material characterization, is usually modelled as a circular retarder followed by a linear retarder. However, this modelThe Mueller matrix polar decomposition method decomposes a Mueller matrix into a diattenuator, a retarder, and a depolarizer. Among these elements, the retarder, which plays a key role in medical and material characterization, is usually modelled as a circular retarder followed by a linear retarder. However, this model may not accurately reflect the actual structure of the retarder in certain cases as many practical retarders do not have a layered structure or consist of multiple (unknown) layers. Misinterpretation, therefore, may occur when the actual structure differs from the model. Here, we circumvent this limitation by proposing to use an elliptical retarder parameter set that includes the axis orientation angle φ, the degree of ellipticity χ, and the elliptical retardance ρ. By working with this set of parameters, an overall characterization of any retarder is provided, encompassing its full optical response without making any assumptions about the structure of the material. In this study, experiments were carried out on liquid crystalline samples to validate the feasibility of our approach, demonstrating that the elliptical retarder parameter set adopted provides a useful tool for a broader range of applications in optical material analysis..
Advanced Photonics Nexus
- Publication Date: Dec. 01, 2025
- Vol. 4, Issue 6, 066015 (2025)
Hybrid physics-informed and data-driven mode solver for optical fiber design | Editors' Pick
Xiao Luo, Min Zhang, Zhuo Wang, Xiaotian Jiang, Yuchen Song, and Danshi Wang
An efficient neural mode-solving operator is proposed for evaluating the propagation properties of optical fibers. By incorporating the governing Helmholtz equation into training, the working mechanism of the proposed operator adheres to the physics essence of fiber analysis. The training of the mode-solving operator aAn efficient neural mode-solving operator is proposed for evaluating the propagation properties of optical fibers. By incorporating the governing Helmholtz equation into training, the working mechanism of the proposed operator adheres to the physics essence of fiber analysis. The training of the mode-solving operator adopts a hybrid physics-informed and data-driven approach, providing the advantages of strong physical consistency, enhanced prediction accuracy, and reduced data dependency in comparison with purely data-driven methods. Benefiting from the improvements in network input–output mapping formulation, the proposed operator offers broader applicability to different fiber types and greater flexibility for property optimization. Combined with the particle swarm optimization and refractive index optimization, the operator demonstrates its capacity for the inverse design of multi-step-index fibers (MSIFs) and graded-index fibers (GRIFs). For MSIFs, to ensure a low mode crosstalk for short-distance transmission systems, optimized refractive index profiles (RIPs) of both three-ring and four-ring structures are obtained from large structure parameter search spaces. For GRIFs, to ensure a low receiving complexity for long-haul transmission systems, optimized RIP with low root mean square mode group delay is obtained through point-wise fine-tuning. Moreover, the operator is capable of analyzing the effect of dopant diffusion in manufacturing..
Advanced Photonics Nexus
- Publication Date: Nov. 20, 2025
- Vol. 4, Issue 6, 066016 (2025)
Physics-informed meta neural representation for high-fidelity, aberration-corrected, sparse-view Fourier ptychographic tomography
Minglu Sun, Fenghe Zhong, Shiqi Mao, Ying Liu, Zihao Zhang, Dongyu Li, Binbing Liu, and Peng Fei
Label-free 3D tomography has attracted growing attention in biological imaging due to its inherent resistance to phototoxicity and concise system configuration. Among existing techniques, Fourier ptychographic tomography (FPT) stands out for high-resolution refractive index (RI) reconstruction from noninterferometric mLabel-free 3D tomography has attracted growing attention in biological imaging due to its inherent resistance to phototoxicity and concise system configuration. Among existing techniques, Fourier ptychographic tomography (FPT) stands out for high-resolution refractive index (RI) reconstruction from noninterferometric measurements, avoiding coherent noise and phase instability—key limitations of optical diffraction tomography. However, conventional FPT suffers from significant artifacts and high computational demands, especially for multiscattering samples and long-term observation. Here, we introduce physics-informed aberration-corrected meta neural representation (PAMR), an advanced self-supervised framework that integrates neural representation with physics prior, meta-learning optimization, and adaptive aberration correction. Simulations and experiments show that PAMR produces high-fidelity 3D reconstructions with reduced artifacts and strong optical section ability, achieving 137 and 550 nm resolution for lateral and axial, respectively. Moreover, PAMR exhibits superior sparse-view robustness, sustaining high-quality with 75% view reduction. Through the meta-learning strategy, the reconstruction speed of dynamic volumes could be increased by 10 times. Applications include 3D RI imaging of multiscattering C. elegans and long-term 3D observation of HeLa cells, showing detailed organelle structures and interactions. As a generalizable approach combining computational efficiency with physical accuracy, PAMR provides an advanced algorithm for label-free 3D microscopy, with broad applicability across biomedical research..
Advanced Photonics Nexus
- Publication Date: Dec. 09, 2025
- Vol. 4, Issue 6, 066017 (2025)
Silicon thermo-optic phase shifters: a review of configurations and optimization strategies
Vol. 3, Issue 4, 044001 (2024)
Flexible depth-of-focus, depth-invariant resolution photoacoustic microscopy with Airy beam
Vol. 3, Issue 4, 046001 (2024)
Highly sensitive mid-infrared upconversion detection based on external-cavity pump enhancement
Vol. 3, Issue 4, 046002 (2024)
Vol. 2, Issue 6, 065001 (2023)
Decision-making and control with diffractive optical networks
Vol. 3, Issue 4, 046003 (2024)
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20