Biomedical Optics Express

Polarization-resolved second harmonic generation microscopy provides structural information about non-centrosymmetric biological samples such as collagen. It involves illuminating the sample with a focused laser beam having a variable linear polarization angle and recording the second harmonic signal as a function of this angle. However, accurate linear polarization control is challenging due to ellipticity introduced by reflections from mirrors and dichroic mirrors in the optical path. Waveplate-based compensation has emerged as the standard approach to address these distortions, but its effectiveness for quantitative measurements remains incompletely characterized. Here, we attempt to fill this gap by implementing an established automated waveplate compensation method based on a rotating half-waveplate in combination with a compensating quarter-waveplate. This was done on a commercial Leica TCS SP8 MP multiphoton microscope, making various hardware improvements and carefully documenting important experimental details. Despite significant effort, we consistently observed substantial unwanted residual polarization ellipticity, with amplitudes up to 0.25, persisting under optimal waveplate configurations. Our simulation analysis provides evidence that this limitation may arise from wavelength-dependent dichroic mirror birefringence combined with the broad spectral bandwidth (10nm to 20nm full width at half maximum) of femtosecond laser pulses. While the approach investigated here can compensate a single wavelength, different spectral components within the pulse experience different phase retardations from wavelength-dependent optical elements, potentially resulting in residual ellipticity that cannot be eliminated. Our simulations qualitatively reproduced key features of the experimental observations. These findings have important implications for quantitative polarization-resolved second harmonic generation microscopy and suggest that alternative approaches, including specimen rotation or picosecond laser sources with narrower bandwidth, should be investigated for applications requiring precise polarization control. To facilitate community investigation of these effects, we provide open-source analysis code and simulation files.

Collagen remodeling in the uterine cervix is a vital process in pregnancy that allows for timely fetal delivery, yet its spatio-temporal details are still not fully understood. In this study, we measured collagen reorganization at different stages of murine gestation and at various cervical depths. We used polarization-resolved Second Harmonic Generation microscopy to specifically detect fibrillar collagen and assess its orientation with sub-micrometer resolution. We imaged large cervical areas using automated mosaicking and implemented an analysis pipeline that showed significant region-dependent changes in collagen quantity, porosity, and orientation disorder. Notably, we found that collagen disorganization begins in the lower cervix at gestation day 12 and extends throughout the entire cervix by day 15. Additionally, we demonstrated that the temporal dynamics of disorganization, without spatial sensitivity, can also be tracked using Mueller Matrix imaging, which is a clinically deployable method. These findings should improve understanding and diagnosis of gestation-related issues such as premature birth.

SignificanceLaser speckle contrast imaging (LSCI) is widely used to measure blood flow, but speckle fluctuations may also encode biologically meaningful dynamics beyond perfusion. Foundational studies in dynamic light scattering (DLS) and micro-optical coherence tomography (OCT) have also demonstrated that slow coherent signal fluctuations can arise from energy-dependent intracellular motion in in vitro and ex vivo systems. Building upon these advances, recent work has shown that LSCI has the potential to detect slow speckle dynamics (SSD) correlated with cellular dynamics in vivo. However, the biophysical mechanisms underlying SSD in intact brain tissues remain insufficiently validated. Establishing a mechanistic bridge from controlled ex vivo and in vitro conditions to in vivo brain measurements is critical for translating speckle-based imaging beyond perfusion measurements to enable label-free assessment of cellular and metabolic activity in disease models. AimThe objective of this study is to investigate the biophysical origin of the SSD in vivo and evaluate its sensitivity to intracellular metabolic activity in brain tissue. ApproachWe utilize an epi-illumination LSCI system to measure speckle contrast as a function of camera exposure time and extract characteristic decorrelation time constants. SSD was investigated in acute mouse brain slices, where blood flow is absent, to eliminate vascular confounds. Cellular metabolism was systematically modulated using 2-deoxyglucose and glucose. Complementary in vivo measurements were performed to reveal SSDs response to hyperoxia and normoxia after ischemic stroke. ResultsSSD signals persisted in acute brain slices in the absence of blood flow. Inhibition of glycolysis significantly reduced SSD, while restoration of metabolic substrates partially recovered the signal. In in vivo measurements, SSD increased during hyperoxia compared to normoxia after ischemic stroke, suggesting increased oxygen-supported cellular metabolic activity. ConclusionsThese results indicate that SSD is sensitive to energy-dependent cellular processes closely tied to metabolic activity. SSD represents a previously uncharacterized, label-free in vivo optical contrast that enables assessment of cellular metabolic activity as well as vascular dynamics. This work establishes a mechanistic foundation for using SSD as a general optical marker of cellular viability in in vivo measurements.

Optical coherence tomography (OCT) has transformed clinical eye care by providing high-resolution volumetric imaging of the retina. Recently, ultrawide-field-of-view (FOV) OCT played an increasingly significant clinical role; however, most clinical OCT systems offer only a rather limited FOV. We increased the FOV of clinical OCT by volumetrically montaging multiple OCT datasets in three dimensions (3D). We performed volumetric montaging by representing the internal limiting membrane (ILM) and retinal pigment epithelium (RPE) in each volume as point clouds and using these point clouds to compute transformations that map each volume to a common reference frame. We validated our methodology using datasets from three institutions with different OCT hardware and data-acquisition procedures. Using the mean surface distance between point clouds, we found the error in montaging was less than the lateral pixel size. Our method enabled existing clinical OCT to achieve ultrawide FOV imaging without any hardware modification.

Accurate cell type classification is essential for a wide range of biomedical applications, including disease diagnosis, drug discovery, and the study of cellular processes. Holographic imaging flow cytometry (HIFC) provides label-free quantitative phase imaging (QPI) of individual cells, enabling classification based on phase images. However, reconstructing holograms into phase images involves multi-step image processing, which introduces substantial computational overhead. The availability of diverse image representations across holographic reconstruction stages allows for flexible analytical strategies, enabling the optimization of trade-off between classification accuracy and computational efficiency. Moreover, deep learning offers an efficient alternative, accelerating the reconstruction process while performing accurate classification. However, despite its importance, this optimization challenge remains largely unexplored in the current literature. Here, we present the first systematic evaluation aimed at balancing classification accuracy with computational efficiency, highlighting how different image representations affect overall performance. We focus on a binary classification task discriminating natural killer cells from breast cancer cells. Six distinct classification pipelines are evaluated: direct processing of raw holograms, analysis of demodulated complex fields (CFs), refocused CFs, unwrapped phase images, and two deep learning-based methods that either replace the automatic refocusing stage or perform end-to-end hologram-to-phase reconstruction. For each strategy, we assess both computational cost and classification performance. Our results reveal a clear trade-off: reconstructed phase images provide the highest accuracy, whereas simpler representations or accelerated reconstruction methods significantly reduce processing time with minimal loss of accuracy. A Pareto analysis identifies the optimal set of strategies, offering practical guidelines for selecting image representations and processing pipelines based on available hardware and desired performance. Thus, this work offers a systematic framework for high-throughput deep learning classification in HIFC, serving as a potential reference for future biomedical applications.

SignificanceCollagen autofluorescence provides valuable intrinsic contrast for assessing tissue structure, composition, and pathology. However, a comprehensive understanding of the fluorescence properties across different collagen types remains limited. This knowledge gap may limit the development of advanced label-free fluorescence spectroscopy and imaging techniques for specific tissue characterization and diagnostic applications. AimThis study aims to comprehensively characterize the fluorescence intensity excitation-emission matrices (I-EEMs) and time-resolved excitation-emission matrices (TR-EEMs) of collagen standards from Types I, II, III, IV, and V obtained from various organ sources under both dry and hydrated conditions, to identify optimal excitation-emission parameters for each collagen type discrimination, and to establish a reference dataset that supports future research in label-free tissue characterization. ApproachWe employed a time-resolved fluorescence spectroscopy system equipped with an optical parametric oscillator laser (excitation: 200-2000 nm, pulse width: 30 ps) as an excitation source to generate I-EEMs and TR-EEMs of human and bovine collagen Types I-V. The fluorescence light was obtained by a multichannel plate photomultiplier tube through a monochromator (spectral range: 200-1000 nm). Measurements were conducted using collagen standards, under both dry and hydrated states. Additionally, photobleaching effects were assessed to ensure the reliability and reproducibility of fluorescence data. ResultsEach collagen type exhibited distinct I-EEM and TR-EEM signatures, with fluorescence lifetimes ranging from 2.5 ns (Type III, bovine skin) to 5.3 ns (Types II and V). Fibrillar collagens (Types I and V) displayed broader I-EEMs, whereas basement membrane collagen (Type IV) showed the narrowest spectral distribution. Organ-source-dependent variations were evident within the same collagen type. Type I collagen from human placenta exhibited an inverse lifetime-emission wavelength relationship compared to bovine sources. Hydration consistently red-shifted emission peaks into the 395-420 nm range and reduced fluorescence lifetimes across all collagen types (e.g., Type I bovine Achilles tendon: 3.2-5.0 ns dry vs. 3.0-4.5 ns hydrated). Despite excitation wavelength- and fluence-dependent photobleaching of fluorescence intensity, fluorescence lifetimes remained relatively stable, confirming the robustness of lifetime-based measurements. ConclusionsThis study establishes a comprehensive reference dataset for the fluorescence properties of collagen Types I-V and demonstrates the potential of combined I-EEMs and TR-EEMs analysis for tissue characterization. The results highlight species-, organ-, type-, and environment-specific optical fingerprints of similar collagens, which must be considered before implementing more in-depth studies on how the optical properties of collagen change in different medical applications.

In contemporary bio-imaging-based research, computer-based assessment is becoming crucial for the characterisation of biological structures, as it minimises the need for time-consuming human annotation, which is prone to human error. Furthermore, it allows for the use of optical techniques that use lower photon intensities, thereby reducing reliance on high-intensity excitation and mitigating adverse effects on their activities. This study details the development and evaluation of sophisticated deep-learning models for amoeba detection using phase-contrast imaging. Using a single-class annotated dataset comprising 88 images and 4,131 annotations, we developed nine object detection models based on Detectron 2 and six variants based on YOLO v10. The diversity of the dataset, acquired under varying setup parameters, facilitated a comprehensive evaluation of the strengths and limitations of each model. A comparative analysis of speed and accuracy was performed to identify the most efficient models for real-time detection, providing critical insights for future microscopic analyses.

BackgroundLive-cell fluorescence microscopy enables the study of dynamic cellular processes. However, fluorescence microscopy can damage cells and disrupt these dynamic processes through photobleaching and phototoxicity. Reducing light exposure mitigates the effects of photobleaching and phototoxicity but results in low signal-to-noise ratio (SNR) images. Deep learning provides a solution for restoring these low-SNR images. However, these deep learning methods require large, representative datasets for training, testing, and benchmarking, as well as substantial GPU memory, particularly for denoising large images. ResultsWe present a new fluorescence microscopy dataset designed to expand the range of imaging conditions and specimens currently available for evaluating denoising methods. The dataset contains 324 paired high/low-SNR images ranging from four to 282 megapixels across 12 sub-datasets that vary in specimen, objective used, staining type, excitation wavelength, and exposure time. The dataset also includes spinning disk confocal microscopy examples and extreme-noise cases. We evaluated three state-of-the-art deep learning denoising models on the dataset: a supervised transformer-based model, a supervised CNN model, and an unsupervised single image model. We also developed an image stitching method that enables large images to be processed in smaller crops and reconstructed. ConclusionsOur dataset provides a diverse benchmark for evaluating deep learning denoising methods, and our stitching method provides a solution to GPU memory constraints encountered when processing large images. Among the evaluated deep learning models, the supervised transformer-based model had the highest denoising performance but required the longest training time.

Longitudinal live cell imaging is valuable for characterizing dynamic morphological and phenotypic changes in biological systems. However, conventional approaches rely on manual microscope operation, which is labor-intensive, limits imaging frequency, and disrupts the cellular environment. These constraints reduce scalability, increase experimental variability, and restrict both the duration and temporal resolution of continuous imaging. Although automated imaging platforms partially address these limitations, existing solutions are often constrained by the cost, footprint, and inflexibility of in-incubator microscopes or stage-top incubators. Here, we present an automated in-incubator epifluorescence microscope designed for long-term operation. The system features a modular architecture with optional multi-fluorescence imaging, automated plate scanning, configurable light sources, and compatibility with multiple plate formats, including integration with fluidic automation devices. By positioning the light sources and control electronics outside the incubator, the platform improves thermal stability and long-term operational reliability. This approach enables continuous, high-frequency imaging over extended durations, providing a source of rich data for quantifying time-dependent tissue phenotypes, morphological remodeling, and transient biological processes.

Oblique plane microscopy (OPM) is a light sheet microscopy technique that uses a single high numerical aperture (NA) objective for both illuminating the sample and collecting emission fluorescence from a tilted plane within the specimen. OPM has become indispensable in biological and biomedical research, providing rapid, high-resolution volumetric fluorescence imaging of live cells and tissues while minimising phototoxicity and photobleaching. It also overcomes the sample mounting challenges associated with conventional light sheet microscopes that require two orthogonally placed objectives. However, the application of OPM has been limited by the complex design and the intricate optical alignment and characterisation needed, particularly with the remote-refocusing system (RFS) in the emission path. This protocol offers a detailed, step-by-step guide for constructing an OPM setup using commercially available components and for characterising its performance to ensure optimal imaging quality. We aim to deliver the unique merits of OPM to researchers in life science and medicine, enabling them to visualise the spatiotemporal organisation of key biomolecules, structures, and cells in 3D at high resolutions.

Cerebral blood volume (CBV) and blood flow (CBF) constitute key metrics for cerebrovascular monitoring, enabling assessment of stroke severity and risk-prediction, aging-related changes, and neurological diseases. CBF and CBV monitoring are key aspects in diagnosis, treatment triage, and clinical outcome of ischemic and hemorrhagic strokes. In recent years, there have been ongoing efforts toward the development of optical devices for noninvasive monitoring of CBV and CBF. Speckle contrast optical spectroscopy (SCOS) has recently emerged as a strong candidate for clinical translation in monitoring CBF and CBV, due to its affordability, compact and wearable design, and noninvasive nature. However, experimental demonstrations that SCOS can effectively monitor brain hemodynamics remain sparse. This is primarily due to challenges in design experiments that isolate cerebral blood dynamics from those in the scalp and skull. In this paper, we report experiments using SCOS to monitor cerebral hemodynamics in rats during intracerebral blood flow modulation. To modify cerebral blood dynamics, a surgical procedure was performed to insert a catheter for direct injection of flow modulation fluids into the brain. Using the SCOS device, we monitored changes in CBV during deliberate CBF interventions into the brains of five rats. A saline solution was also injected as a sham control of the flow intervention. The results show a significant decrease in CBV during injection, followed by a return to baseline. This behavior is consistent with physiological expectations, as the injected fluids dilute the blood, leading to a transient reduction in blood volume. Notably, the CBV decrease induced by the flow modulation fluid solution required more than twice as long to recover to baseline compared with the saline solution, which is consistent with the delayed clearance of the flow modulation fluid by design. These experimental results demonstrate the effectiveness of SCOS for monitoring cerebral hemodynamics in animal models and highlight its potential for translation to human studies. Moreover, this work paves the way for the testing and characterization of cerebral therapeutic agents intended for blood flow modulation in animal models.

Structured Illumination Microscopy (SIM) provides imaging with spatial super-resolution, as well as optical sectioning capability, without relying on specialized fluorescent dyes. 2D and 3D variants of this method exist, but most bespoke implementations are 2D-SIM, because it is easier to realize and modify than 3D-SIM. 2D-SIM systems, however, often experience reconstruction artifacts, especially when pushing for high lateral spatial resolution in thicker samples. We present enhanced 2D-SIM, an approach to 2D-SIM where both, coarse patterns optimized for removing out-of-focus background, and fine patterns optimized for resolution improvement beyond the diffraction limit are used. In combination, this achieves 2D-SIM reconstructions with high contrast, spatial super-resolution, and significantly reduced reconstruction artifacts. We present the theoretical framework of this technique, and provide enhanced 2D-SIM imaging results of liver sinusoidal endothelial cells stained with fluorophores emitting at visible and near-infrared wavelengths. Quantitative comparisons of power spectral distribution and image resolution are provided.

Digital holographic microscopy systems in a common-path configuration, compared to systems with a separate reference arm, offer a compact design and resistance to disturbances. They can operate with partially coherent illumination, reducing speckle noise. However, they are limited by the overlapping of the object beam and its laterally shifted replica. As a result, images from different regions of the object overlap on the detector, preventing imaging of dense samples. We present the wavelength-scanning replica-removal method, which solves this problem by enabling the separation of information from both replicas and thereby doubling the effective field of view (FOV). The wavelength-scanning multi-shear replica removal algorithm plays a key role in reconstructing the undisturbed phase from a series of holograms recorded with variable shears. The shear value is controlled by changing the illumination wavelength. This enabled the development of two measurement modes: time-domain wavelength scanning for high-quality imaging, and a single-shot mode with frame division into color channels to improve temporal resolution. The method was validated using resolution tests and biological samples - neurons and dynamic yeast cultures. By combining the advantages of the common-path configuration with dense-structure imaging and dynamic processes, the proposed method constitutes a versatile tool for quantitative phase microscopy.

Mapping peritumoral collagen fiber directionality in solid tumors may assist in determining cancer progression and support more personalized prognoses. However, existing microscopy techniques are often limited by a restricted field of view, high cost, or incompatibility with paraffin-treated tissues. Computational scattered light imaging (ComSLI) is a cost-effective whole-slide microscopy technique that reveals fiber orientations independent of sample preparation. Using glioma, colorectal, and head and neck cancer samples, we show for the first time that ComSLI maps fiber orientations in paraffin-treated tumor tissues, visualizes tumor growth pathways and desmoplastic reactions, and allows the study of collagen orientations relative to tumor boundaries.

Spectral information plays a crucial role in biological imaging, yet conventional epifluorescence and histological techniques often rely on RGB image acquisition, limiting the resolution of spectrally overlapping components. Here, we present a phasor-based spectral analysis framework adapted for RGB images, enabling unsupervised segmentation and unmixing without the need for hyperspectral systems or sequential acquisition. By applying a discrete Fourier transform to the red, green, and blue intensities at each pixel, we generate a two-dimensional phasor plot where spectral relationships are encoded in modulation and phase. We demonstrate the utility of this approach across three distinct applications: segmentation of lung histology images stained with hematoxylin and eosin to quantify alveolar collapse, analysis of autofluorescence in skin lesions (nevi and melanoma) to highlight pathological spectral signatures, and spectral unmixing in multicolor-labeled U2OS cells to resolve overlapping fluorophores. Our method improves signal separation, reduces noise, and enhances biological interpretability using standard RGB acquisition. These findings establish RGB phasor analysis as a practical and powerful tool for spectral decomposition and segmentation in microscopy, bridging the gap between conventional imaging and advanced spectral analysis.

This study uses two different quantitative phase imaging techniques (QPI) and for the first time measures the height, volume, and mass dynamics of Madin-Darby Canine Kidney (MDCK) monolayers. We demonstrate novel methods to determine the height of confluent monolayers of cells from 2D and 3D QPI data and validate that the two methods agree. We developed a novel cell segmentation method adapted to QPI images of confluent cell layers and present a robust measure of relative error. We also demonstrate that height statistics of cells can be obtained without segmenting the images. We obtain the following precisions of cell density (1 %), height (3 %), area (5 %) and volume (6 %). Cell height varies 15-25 % over a monolayer and increases 50-100 % when cell density doubles. The average refractive index and the dry mass fraction of the cells, on the other hand, are constant over the entire density range.

The ability to detect host cell factors during Staphylococcus aureus infection in vitro by immunofluorescence microscopy is severely hampered by staphylococcal protein A (SpA), a cell wall-anchored protein that binds the fragment crystallizable (Fc) region of immunoglobulins. This interaction generates strong nonspecific fluorescent signals on the bacterial surface, complicating data interpretation and limiting the accuracy of quantitative image analysis. Several measures have been put forward to overcome this obstacle, most importantly the pre-incubation with an anti-SpA antibody (SpA) and the use of human serum (HS) as blocking agent and antibody diluent. To highlight this feature to general fluorescence microscopy users, we here systematically evaluated these two strategies. Using S. aureus coated on coverslips and S. aureus-infected A549 cells, we highlight the efficiencies of both approaches to markedly reduce nonspecific fluorescence, with HS treatment yielding the most profound suppression. Notably, HS, containing high levels of human immunoglobulins, offered a robust, cost-effective and broadly applicable solution for minimizing SpA-driven artifacts, thereby improving immunofluorescence microscopy in S. aureus infection models in vitro.

The performance of a laser scanning microscope inevitably depends on the performance of the point detector. As laser scanning approaches aim to penetrate deeper in tissue, there is a commensurate need for detectors that can operate with high sensitivity, bandwidth, and dynamic range at near-infrared wavelengths where scattering is reduced. Here, we demonstrate that fiber optical parametric amplification can be used to boost low-power microscopy signals to levels that can be detected by near-infrared photodiodes without introducing prohibitive noise. We construct amplifiers that achieve >50 dB of parametric gain at wavelengths within the third near-infrared transparency window and have similar sensitivity to near-infrared photomultiplier tubes. Furthermore, these amplifiers outperform detection with a photodiode and subsequent electrical amplification, providing a factor of 10-100-fold improvement in sensitivity. We demonstrate amplifier bandwidths up to ~1.6 GHz, a factor of 10 faster than conventional detectors, including near-infrared photo-multiplier tubes, with sensitivity of ~8 nW (corresponding to ~20 photons/pixel). Finally, the increased performance of the optical amplifier is confirmed in diagnostic imaging experiments where >10x less power is required to achieve the same signal-to-noise ratio and contrast as images using electrical amplification. Accordingly, fiber optical parametric amplification is a new path forward for extending the performance of laser scanning microscopes in the near infrared.

Microscopy images are frequently downsampled due to acquisition and computational constraints, requiring reconstruction before downstream analysis. While super-resolution (SR) is typically assessed using pixel-level fidelity metrics, its impact on deep learning (DL) model behavior remains insufficiently understood. In this work, we present a study that examines how different upsampling strategies affect image quality and classification performance. Using the BloodMNIST dataset, we construct matched 224x224 datasets from 64x64 images via bicubic interpolation, SwinIR Classical, and SwinIR RealGAN DL SR models, alongside the original 224 ground-truth images. We evaluate reconstruction quality using the Structural Similarity Index Measure (SSIM) and Peak Signal-to-Noise Ratio (PSNR) scores and assess downstream classification performance using ResNet-50 and Vision Transformer models, with accuracy, macro-F1 score, and a confidence-aware metric, the area under the receiver operating curve for successful prediction (AUPR Success). Our results demonstrate that bicubic interpolation significantly degrades classification performance, whereas SR methods can recover class-relevant information, even better than the ground-truth data. These findings emphasize the importance of confidence-aware evaluation and unambiguous reporting of reconstruction pipelines in microscopy-based DL studies.

A wide variety of protocols have been proposed for optical clearing of tissues, whole-mount organs, and other bulky specimens to enable their volumetric fluorescence imaging. However, quantitative comparisons of tissue clearing protocols that take into account the fluorescence of the final specimens remain rare. Here, we propose a volumetric fluorescence image-based workflow for evaluating tissue clearing and fluorescence staining protocols. The workflow calculates depth-dependent fluorescence attenuation coefficients using data from entire 3D images, thereby avoiding spatial sampling bias and eliminating reliance on simple readouts, such as light transmittance, to predict fluorescence image quality. By combining autofluorescence signal with the signal from a specific fluorescence label, we independently evaluated transparency and the quality of fluorescence staining in cleared specimens. Using the proposed workflow, we systematically compared clearing and staining performance of three CUBIC-based protocols in murine liver, kidney, spleen, thymus, and intestine, and revealed differences in final fluorescence image quality across protocol-organ combinations.