Histological staining remains the diagnostic gold standard in renal pathology, but it requires labor-intensive preparation, specialized infrastructure, and skilled personnel. Isotropic quantitative differential phase contrast (iDPC) microscopy provides label-free morphological contrast by recovering the optical phase of transparent specimens under asymmetric illumination. In this work, we extend label-free virtual histological staining into an end-to-end computational pathology pipeline: iDPC phase maps are first translated into virtually stained bright-field images using an unsupervised image-to-image translation framework without paired training data, and the resulting images are then used for downstream automatic glomerulus segmentation to localize glomerular regions and subsequently classify each segmented glomerulus as normal or abnormal, enabling quantitative, region-focused screening on unstained renal tissue sections.

Abstract Three-dimensional (3D) tracking of microscopic objects is essential for probing dynamic processes across biological, microfluidic, and nanoscale systems. In-line digital holographic microscopy (DHM) provides a powerful framework for this purpose by encoding volumetric information into a single intensity measurement, enabling scan-free reconstruction through numerical wave propagation. However, despite its conceptual simplicity, the practical implementation of in-line DHM remains fundamentally constrained by its reliance on bulky relay optics, which limit system compactness, alignment robustness, and integration with emerging miniaturized platforms. Here, we rearchitect in-line DHM using planar meta-optics by replacing conventional refractive relay optics with metalenses, thereby establishing a compact single-path holographic imaging platform. This architecture preserves the intrinsic advantages of in-line holography, including single-shot volumetric encoding and numerically propagated reconstruction, while substantially reducing optical complexity and system footprint. We validate the metasurface-enabled in-line DHM experimentally using resolution targets, microspheres, and live microorganisms and demonstrate scan-free 3D tracking in a compact optical configuration. Our results establish planar meta-optics not only as a substitute for refractive components but also as an effective system-level strategy for miniaturizing holographic microscopy and advancing integrable volumetric imaging platforms.

We present a compact metasurface-based Fourier ptychographic microscope integrating metalens optics and programmable illumination to achieve miniaturized, wide-field quantitative phase imaging with nearly twofold resolution improvement and demonstrated performance on biological samples.

qDPC microscopy enables rapid acquisition of quantitative phase images for improved sample characterization. However, conventional methods require careful regularization tuning. Here, we propose a self-supervised approach for biological samples, enabling robust reconstruction and virtual staining for biomedical applications.

In this work, we utilized an unsupervised CycleGAN framework to generate HiLo-equivalent images from single fiber-bundle acquisitions, eliminating structured illumination while improving optical sectioning and enabling faster, real-time, clinically practical minimally invasive imaging.

Photoacoustic microscopy (PAM) is a promising imaging technique that combines optical excitation with ultrasonic detection. However, most PAM systems operate in the visible wavelength range and in transmission configuration, which limits the types of samples that can be imaged and provides mainly appearance information. By designing a metalens for the mid-infrared region, it becomes possible to develop a more flexible and compact reflection-mode PAM system for label-free dynamic imaging of living cells.

Light-sheet fluorescence microscopy enables fast volumetric imaging with minimal photobleaching and phototoxicity. To extend the field of view while simplifying the experimental setup, we use a cubic phase metasurface to generate Airy light-sheet illumination.

The Optica Digital Holography and Three-Dimensional Imaging Meeting (DH), held as part of the Imaging Systems and Applications Congress in Seattle in August 2025, served as an important forum for international researchers to present recent work and discuss developments in digital holography, three-dimensional imaging, and optical information processing. The DH meeting provides a venue for research on the science, technology, and applications of digital holography, 3D imaging, and optical information processing methods.

Accurate assessment of mucosal inflammation is critical for monitoring ulcerative colitis severity, yet manual Mayo Endoscopic Subscore (MES) classification remains subjective and time-consuming. To address these challenges, we propose an interpretable deep learning framework for preliminary MES classification on colonoscopy data. We utilize a Modified Classification Special Euclidean (Mod-Cls-SE(2)), designed to handle rotation variability in colonoscopy imaging. The architecture integrates handcrafted texture features with geometric deep learning and a lightweight feedback agent. Texture features are extracted via gray-level co-occurrence matrix (GLCM) statistics and wavelet energy descriptors. These features are projected into a low-dimensional embedding using Uniform Manifold Approximation and Projection (UMAP) to enable interpretable rule-based logic. The resulting representation is fused with features learned by a customized Mod-Cls-SE(2) backbone tailored for endoscopic imaging. Unlike conventional networks, which are sensitive to camera orientation, our Mod-Cls-SE(2) model operates in roto-translation space, ensuring consistent feature extraction across varying viewing angles. To enhance adaptability, we introduce a feedback agent based on LightGBM, which is iteratively trained on misclassified samples from the dataset. This agent refines classification decisions without requiring full retraining of the core network. Experiments conducted on a curated public colonoscopy dataset demonstrate that our pipeline improves MES classification accuracy from 69.7% to 87.1% after several feedback iterations. This preliminary study highlights the potential of a feedback-guided, geometry-aware framework for interpretable and robust endoscopic inflammation scoring. Future work will focus on validating the system framework to assess real-world applicability and performance.

We demonstrate a PQ:PMMA volume holographic lenslet array illuminator that generates uniform super-Gaussian multifocal spots for faster CCD-based multifocal confocal imaging with digital pinholing and image stitching, while maintaining optical sectioning and image quality.

Autofocusing circular Airy beams and their superposition’s were numerically studied using the vector angular spectrum method. Radial polarization enhanced longitudinal-field-mediated compression, producing lens-free needle-like foci. Results provide insights for structured-light engineering and fluorescence microscopy applications.

We present an unsupervised RCAN-CycleGAN framework that translates single-shot wide-field fluorescence endomicroscopy images into high-contrast, optically sectioned HiLo images using only unpaired training data. Trained on unpaired wide-field and HiLo images from fluorescent beads, ex vivo mouse brain, and plant specimens, the model generates HiLo-quality output from a single wide-field exposure without structured illumination hardware.

Compact photoacoustic microscopy design remains challenging due to optical component complexity. To address this challenge, we demonstrate the use of a 3-mm-diameter metalens integrated with an ultrasound transducer, enabling A-scan detection and B-mode imaging.

This study benchmarks colonoscopy image reconstruction. We introduce a Latent Bank method to improve standard encoders, but High-Fidelity GAN Inversion achieves superior quality (FID = 22.12, MS-SSIM > 0.91), proving the best for colonoscopy images.

Optical sectioning endomicroscopy significantly improves image contrast by rejecting out-of-focus background fluorescence, but conventional HiLo microscopy requires at least two structured-illumination exposures and, in some cases, axial scanning, which increases acquisition time and system complexity. Obtaining large, perfectly registered wide-field and HiLo image pairs for supervised deep learning is also extremely challenging in vivo. Here, we present an unsupervised RCAN-CycleGAN framework that translates single-shot wide-field fluorescence endomicroscopy images into high-contrast, optically sectioned HiLo images using only unpaired training data. Trained on unpaired wide-field and HiLo images from fluorescent beads, ex vivo mouse brain, and plant specimens, the model generates HiLo-quality output from a single wide-field exposure without structured illumination hardware. On an independent test set of paired images, our method achieves an average PSNR of 32.2 dB and SSIM of 0.9 with respect to experimental HiLo images, gains of >13 dB in PSNR and >0.7 in SSIM over raw wide-field inputs. Compared with other methods on the same dataset, the proposed RCAN-CycleGAN consistently outperforms the unsupervised baseline CUT as well as CycleGAN variants using U-Net generators. Notably, it also shows better generalization on previously unseen in vivo mouse brain data than the supervised Pix2Pix baseline. These results demonstrate that high-fidelity optical sectioning can be achieved computationally from conventional wide-field endomicroscopy in a fully unpaired manner, enabling a compact, real-time, single-shot solution that can facilitate optically sectioned imaging in clinical and resource-limited settings.

Light-sheet fluorescence microscopy enables optically sectioned images for volumetric imaging. To simplify the illumination arm configuration while minimizing sample-induced stripe artifacts, we use metalens-based dual-sided illumination combined with deconvolution to further improve image quality.

This joint feature issue of Optics Express, Applied Optics, and Biomedical Optics Express showcases recent work from the Computational Optical Sensing and Imaging community. The issue reflects the continuing expansion of computational optical sensing and imaging, where optical design, detection, calibration, physical optical system modeling, signal processing, and machine learning to acquire task-relevant information. The selected papers span computational microscopy and biomedical imaging, adaptive optics, wavefront sensing, holography, phase retrieval, single-pixel and ghost imaging, hyperspectral and polarization imaging, non-line-of-sight imaging, LiDAR, photon-efficient sensing, optical-system co-design, and emerging applications in medicine, remote sensing, manufacturing, and environmental metrology. Together, these contributions illustrate how computationally co-designed optical systems are moving from algorithmic demonstrations toward robust, application-specific imaging and sensing platforms.

Accurate mucosal inflammation assessment in ulcerative colitis is addressed through Mod-SE(2), handcrafted features, UMAP embeddings, and a LightGBM feedback agent. Results show 89.01% multi-domain accuracy on average, demonstrating robust, resilient, generalizable Mayo Endoscopic Subscore classification.

We developed two metalens objective designs to explicitly assess aberration correction: an astigmatism-corrected objective and an otherwise matched uncorrected reference. The corrective phase profile was obtained through ray-tracing–based optimization, and manufacturability was ensured using RCWA-derived nanopillar lookup libraries for layout generation. Beam-profile characterization quantitatively verified aberration suppression, showing a more isotropic PSF and improved, symmetric lateral confinement.