Quantitative analysis of viscoelastic alterations in oral submucous fibrosis (OSF) provides crucial insights for monitoring disease progression and preventing malignant transformation. We developed a piezoelectric-based optical coherence elastography (OCE) system for real-time, quantitative assessment of OSF progression. In our experimental model, sixteen rats were systematically divided into four groups representing progressive fibrosis stages. Phase-sensitive OCE measurements captured distinctive elastic wave propagation patterns across all experimental groups. Comprehensive analysis of phase velocity dispersion curves and wave attenuation enabled the extraction of quantitative viscoelastic parameters that reflect fundamental tissue changes. Results demonstrated significant biomechanical alterations with disease progression, most notably a nearly four-fold increase in Young's modulus from normal tissue (32.6 ± 3.9 kPa) to severe fibrosis (121.1 ± 9.9 kPa), accompanied by more than doubled viscosity coefficients (0.52 ± 0.06 Pa·s to 1.27 ± 0.15 Pa·s). Particularly significant was the loss factor (G"/G') pattern, which exhibited a non-monotonic trend-decreasing from 0.30 in control specimens to 0.18 in moderate fibrosis groups before slightly increasing to 0.20 in severe fibrosis groups. The viscoelastic parameters quantified by OCE may facilitate more precise staging of OSF and potentially provide early indicators for assessing progression risk toward malignancy.

Slide-free microscopy methods for histological imaging in thick, minimally processed tissue specimens are rapidly emerging for surgical margin analysis applications. Here, we report an optical imaging system that combines 266 nm excited fluorescence, 266 nm confocal reflectance, and 1050 nm spectral-domain optical coherence tomography. The deep UV confocal reflectance and DAPI stain emission contrasts provide histology-like imaging at the tissue surface, with virtual H&E staining enabled by deep learning. Cross-sectional OCT extends visualization of tissue morphology beneath the specimen surface with near-infrared depth penetration. A custom reflective objective facilitates simultaneous high resolution, aberration-free imaging of UV confocal reflectance and fluorescence emission with a 0.52 numerical aperture, while transmitting a secondary low numerical aperture OCT beam. Imaging of ex vivo specimens is demonstrated in murine kidney and liver, and human breast tissues.

Drusen are a hallmark biomarker of age-related macular degeneration (AMD), with their size, number, and morphology (type) closely linked to disease severity and progression. Accurate segmentation and classification of drusen from optical coherence tomography (OCT) images are essential for objective AMD assessment and monitoring. In this work, we present a deep learning framework that combines a convolutional neural network for automated drusen segmentation with a dedicated classification module to distinguish four clinically relevant, distinct drusen types based on segmentation output. We evaluated our approach on a comprehensive dataset and achieved a mean Dice score of 0.74 ± 0.21 for voxel-wise segmentation accuracy and a critical success index of 0.69 ± 0.24 for drusen count accuracy. This method demonstrates substantial improvements in the quantitative drusen analysis and offers a promising tool for enhanced AMD diagnosis and tracking of disease progression.

This study presents a hybrid deep learning (DL) approach for designing and optimizing a photonic crystal fiber (PCF)-based surface plasmon resonance (SPR) multi-analyte biosensor. Using a simulation based on the finite element method (FEM), we generated a comprehensive data set that captures various sensor parameters and refractive index (RI) values. A hybrid recurrent neural network long-short-term memory (RNN-LSTM) model was developed to predict confinement loss (CL), which showed superior performance with an MSE of 0.0014, an MAE of 0.0188, and an R of 0.9510 compared to other DL and machine learning (ML) models. The proposed model shows a maximum amplitude sensitivity (AS) of 3102.41 RIU, a wavelength sensitivity (WS) of 10,000 nm/RIU, and a sensor resolution (SR) of 1 × 10. The effectiveness of the model was validated through extensive analysis, including ablation studies and SHAP-based explainability analysis. Our findings highlight the potential of DL to improve multi-analyte biosensor design and performance prediction.

Cellular reactive oxygen species (ROS), a key parameter involved in cell metabolism, signaling, and apoptosis, whose detection is necessary to achieve in a variety of biological processes. However, current ROS detection methods, including fluorescence, colorimetry, and electrochemical methods, are difficult to achieve in-situ non-invasive detection due to their reliance on invasive probes or destructive sampling. In this study, we propose an in-situ non-invasive ROS detection integrating the Raman spectrum and a bidirectional gated recurrent unit (Bi-GRU) deep learning model during HepG2 cell apoptosis. The Bi-GRU model leverages bidirectional gating mechanisms to capture long-term dependencies in Raman spectra while incorporating both forward and backward spectral information for enhanced feature extraction. After training with spectral data of HepG2 cells in various apoptotic states, the R (coefficient of determination) of the Bi-GRU model reaches 0.8511, which outperforms that of traditional methods such as KNN (0.2607), PLS (0.4720), and RNN (0.7724). In the present study, we not only realized the in-situ and non-invasive cellular ROS detection but also expanded the application of artificial intelligence in the field of cellular medicine. Importantly, this will provide a new research idea for further understanding the physiological state of cells and the mechanism of drug action.

Light sheet fluorescence microscopy (LSFM) has transformed the way we visualize biological tissues in three dimensions, offering high-resolution imaging while minimizing photo-induced damage to the samples. Recent breakthroughs in tissue-clearing methods have further improved LSFM's capabilities, making it possible to study larger, intact samples in unprecedented detail. To overcome limitations like shallow penetration and light diffraction in traditional LSFM setups, advanced beam shaping with devices such as spatial light modulators and digital micromirror arrays has been utilized. These improve the resolution and the extent of tissues that can be imaged. Advances such as Bessel-beam-based LSFM and lattice light sheet microscopy increase the field of view that can be imaged with low background noise, but often require complex and bulky equipment. Addressing these complexities, a new approach builds on silicon nitride photonic integrated circuits to create a structured light sheet in a very compact device. This system incorporates beam-steering based on wavelength control in a limited tuning range and enables the generation of light sheets with optimized characteristics in regard to thickness and diffraction-length-limited penetration depth. Simulations include extensive fabrication tolerance analysis that confirms the practicability of the approach, which can be straightforwardly extended to dual wavelength excitation. This compact, chip-based LSFM system could make high-quality imaging more accessible and transform biomedical instrumentation.

Vascular-targeted photodynamic therapy (V-PDT) is a promising treatment for benign vascular proliferative disorders. However, its efficacy largely depends on clinicians' experience due to the lack of reliable methods for efficacy prediction. To provide an objective prediction approach, a hyperspectral imaging (HSI) system was developed to achieve real-time, non-invasive visualization of V-PDT dose parameters, including photosensitizer distribution, oxygen concentration, and vasoconstriction. Based on these measurements, we proposed a photodynamic therapy efficacy prediction index (PEPI)-a new metric that integrates the dynamic changes of both photosensitizer and oxygen throughout the treatment process, thereby providing critical insights for optimizing treatment protocols. Experimental results obtained using a dorsal skinfold window model demonstrate that the system accurately detects V-PDT dose parameters, and the proposed efficacy prediction parameters exhibit a strong positive correlation with treatment outcomes. This work highlights the potential of hyperspectral imaging to advance V-PDT toward more precise, individualized, and effective clinical applications, paving the way for its broader adoption in the field of precision medicine.

Recent advancements in multispectral (MS) and hyperspectral (HS) microscopy have focused on sensor and system improvements, yet sample processing remains overlooked. We conducted an analysis of the literature, revealing that 40% of studies do not report sample thickness. Among those that did report it, the vast majority, 98%, used 2-10 µm samples. This study investigates the impact of unstained sample thickness on MS/HS image quality through light transport simulations. Monte Carlo simulations were conducted on various tissue types (i.e., breast, colorectal, liver, and lung) using optical property parameters extracted from the literature. The simulations revealed that thin samples reduce tissue differentiation, while higher thicknesses (approximately 500 µm) improve discrimination, though at the cost of reduced light intensity. Although the results are based on idealized conditions and exclude certain real-world factors such as sample variability and instrument-specific effects, they highlight the need to study and optimize sample thickness for enhanced tissue characterization and diagnostic accuracy in MS/HS microscopy.

Optical monitoring of cardiac pulsations using near-infrared spectroscopy (NIRS), photoplethysmography (PPG), and diffuse correlation spectroscopy (DCS) is often hindered by motion artifacts and noise. We introduce a synthetic-data-driven framework using a long short-term memory (LSTM) network to trace and denoise pulsatile optical waveforms without reliance on annotated clinical datasets. Physiologically realistic pulsatile signals are generated, corrupted with parameterized artifacts, and used to train the LSTM model. Applied to experimental NIRS, PPG, and DCS signals, the model recovered beat-to-beat morphology more effectively than widely used wavelet and temporal derivative distribution repair (TDDR) filters. Heart rate (HR) extraction from LSTM-processed signals closely matched ECG-derived measurements (mean absolute error = 0.59 bpm, root mean square error = 0.74 bpm). This flexible approach shows potential for rapid adaptation across various devices and noise conditions.

Holographic imaging flow cytometry (HIFC) can generate 2D quantitative phase maps of flowing cells in microchannels. When combined with convolutional neural networks (CNNs), HIFC could provide a promising stain-free approach for identifying target cells in complex cellular environments by leveraging the distinctive morphological and optical properties of different cell types. Here, we propose a lightweight CNN for HIFC image classification, tailored to distinguish ovarian cancer cells from surrounding non-neoplastic cell populations of the tumor microenvironment (TME). We show that the proposed CNN outperforms commonly used models, i.e., Resnet and VGG, with a computational cost lower than Mobilenet, the benchmark for efficiency and accuracy. Our approach could streamline ovarian cancer diagnostics and improve understanding of the TME, ultimately aiding the development of personalized treatments.

Adaptive optics (AO) ophthalmoscopes allow high-resolution imaging of retinal structure and function at the cellular level. Due to their high magnification and small field-of-view (FOV), these systems require precise fixation and light delivery to control the retinal region being imaged and stimulated. We present a high-efficiency fixation and stimulus channel for AO ophthalmoscopy, offering an extended working distance, wide steering range, and broad dioptric correction. For stimulation, the channel delivers intense, near-monochromatic light flashes across much of the visible spectrum. Our design uses all stock components, except for a 3D-printed conic mount and a few machined parts. We balance key system trade-offs and demonstrate design performance through several AO optical coherence tomography (AO-OCT) structural and functional imaging examples. Although originally developed for the Indiana AO-OCT system, these design principles can be readily applied to other AO ophthalmoscopic platforms.

Optical coherence tomography angiography (OCTA) is a pivotal imaging modality for non-invasive visualization of the retinal microvasculature, but current clinical OCTA systems lack the capability to segment and quantify vascular features separately for arteries and veins. This study introduces OCTA-ReVA , an open-source, fully automated toolbox that integrates deep learning-based artery-vein (AV) segmentation and vessel-specific quantitative analysis of OCTA images. OCTA-ReVA computes a comprehensive set of vascular metrics including blood vessel density (BVD), vessel skeleton density (VSD), vessel perimeter index (VPI), blood vessel caliber (BVC), blood vessel tortuosity (BVT), vessel complexity index (VCI), perfusion intensity density (PID), vessel area flux (VAF), and normalized blood flow index (NBFI) independently for arteries and veins. These features are extracted within a user-friendly graphical interface and demonstrate high repeatability and segmentation consistency. By separately quantifying arterial and venous alterations, OCTA-ReVA fills a critical gap in OCTA analytics, enhancing detection and monitoring of retinal vascular diseases.

Surgical resection remains a key curative option for cancer, with minimally invasive approaches increasingly adopted. To enhance intraoperative visualization, we developed a laparoscopic near-infrared hyperspectral imaging (NIR-HSI) system comprising a custom laparoscope, supercontinuum light source, and acousto-optic tunable filter. Ex vivo NIR-HSI of porcine arteries, mesentery, and nerves revealed distinct spectral signatures from 1000-1402 nm. Pixel-based classification via neural networks achieved >99% accuracy, sensitivity, and specificity in most cases. In vivo imaging of a living pig enabled identification of exposed nerves (88.4% accuracy, 68.7% recall) and unexposed arteries (83.2% accuracy, 60.2% recall). These results demonstrate that laparoscopic NIR-HSI can differentiate tissues with similar coloration and detect structures embedded beneath the surface, offering potential for safer minimally invasive surgeries.

Quantitative pathology remains limited due to the need for chemical staining and subjective interpretation of tissue features. Autofluorescence imaging offers a label-free alternative; however, high-dimensional excitation-emission datasets pose challenges for visualization and reproducible analysis. Here, we present Dimensionality Reduction for Enhanced Autofluorescence Microscopy (DREAM), a method that condenses multi-excitation emission spectra into a compact, information-rich format using phasor-based tools. Applied to unstained esophageal tissue samples, DREAM enables high-contrast visualizations that distinguish key histological structures without the need for exogenous labeling. Quantitative assessments across multiple datasets show DREAM improves colorfulness, sharpness, and consistency over single-laser acquisitions, supporting its potential to advance objective, label-free diagnostics through enhanced spectral visualization.

Fluorescence lifetime imaging (FLIm) can detect macroscopic tumor tissue in various organs by measuring tissue autofluorescence, making it a compelling tool for surgical guidance. However, the fluorescence lifetime characteristics of tissue autofluorescence are complex due to the unpredictable microenvironment of the biomolecules in tissue, which complicates data interpretation. Nevertheless, the phasor analysis method is computationally fast and easily interpretable, making it appealing for clinical applications of FLIm. While many implementations of the phasor analysis operate only at a single frequency or a few harmonic frequencies, the phasor theory applied to pulse sampling FLIm as presented in this study leverages the maximum amount of frequency information, thereby extending the set of features available for tissue characterization. The clinical effectiveness of utilizing the maximum range of frequencies in phasor theory applied to pulse-sampling FLIm is demonstrated by investigating tumor detection in ex vivo tissue from 12 patients with prostate cancer. By accounting for the zonal anatomy of the prostate, it is shown that the degree of separability between healthy and tumor tissue is a function of frequency, and hence, the ability to access arbitrary frequency content can improve tumor detection in clinical guidance.

Keratoconus is an ophthalmopathy characterized by a central thinning that commonly causes irregular astigmatism and high myopia. However, the diagnostic standards of keratoconus have not yet been well established clinically and mainly rely on macro characteristics measured through corneal topography. To describe the micro changes conveyed by collagen microstructures in the corneal stroma and provide a new insight for keratoconus, we developed a quantitative diagnostic method combining macro-micro information via optical coherence tomography (OCT). A comparison experiment was designed to confirm the feasibility of characterizing collagen organization using OCT. Macro variables, including heterogeneity in curvature and thickness of cornea, and micro variables of collagen fiber alignment, were calculated and weighted to define the keratoconus potential index (KPI) as a quantitative measure for diagnosing keratoconus and mapping disease risks of 104 participants, which showed excellent diagnostic power (with an area under the curve of 0.991) in keratoconus detection.

Early diagnosis of skin lesions is crucial for improving treatment outcomes. So far, based on optical coherence tomography (OCT), as a non-invasive technique, OCT structural image, optical coherence elastography (OCE), and OCT-based angiography (OCTA) have been utilized to evaluate the biomechanical properties and vascular network in in-vivo human skin. However, image registrations are the major difficulty in separating scans. Therefore, the integration of these three modalities has been achieved by a new scanning protocol to provide a comprehensive, non-invasive assessment of skin tissue, enabling more accurate differentiation between benign and malignant lesions. This study aimed to collect data for the OCT structure, OCE, and OCTA in one scanning acquisition by employing a swept-source (SS-OCT) OCT system. Eleven health participants were recruited for three positions: palm (n = 11), forearm (n = 11), and facial skin (n = 5). From OCE data, Young's modulus was calculated for the stiffness; from OCTA data, vessel area density (VAD), vessel skeleton density (VSD), vessel diameter index (VDI), and weighted Tortuosity Index (WTI) were used to evaluate the vessel network. One facial lesion dataset was collected, and the results indicated differences in the above parameters compared to healthy facial skin data. In conclusion, the new scanning protocol for integrating the structural image, OCE, and OCTA in one scan demonstrated results to calculate the parameters of the skin, which provides potential benefits for skin research and offers full aspects for dermatologists of skin disease.

Vocal fold paralysis (VFP) is characterized by impaired vocal fold movement, commonly resulting from nerve damage during surgical procedures. Current diagnostic methods rely on endoscopic examinations requiring specialized physicians, reducing accessibility and potentially delaying treatment. We propose a non-contact optical sensing method using speckle pattern analysis for VFP identification. Our approach uses external laser illumination and a camera that captures speckle patterns, providing a non-invasive and real-time assessment. The technique uses spectral analysis enhanced by sliding window scanning to extract amplitude peaks across vocal fold regions. Our clinical measurements on 10 subjects (3 healthy and 7 VFP patients) demonstrate identical bilateral voice frequencies, but amplitude varies significantly according to the paralysis side. Healthy subjects presented amplitude ratios close to 1, while VFP patients showed distinct asymmetric patterns: ratios below 0.5 for right-sided paralysis and above 2 for left-sided paralysis, enabling effective VFP detection and localization with potential for clinical implementation.

Photoacoustic imaging (PAI) combines optical contrast with acoustic detection to enable high-resolution, molecular imaging at clinically relevant depths. This review outlines the current status and future potential of contrast-enhanced PAI in human applications. We begin by discussing regulatory considerations surrounding both imaging devices and exogenous contrast agents, highlighting safety concerns, lack of standardized validation protocols, and barriers to the approval of novel agents. To accelerate clinical adoption, many studies have focused on repurposing FDA-approved agents such as indocyanine green, methylene blue, and clofazimine, which offer favorable optical properties and known safety profiles. We then review clinical applications of contrast-enhanced PAI across organ systems. In lymphatic imaging, PAI enables noninvasive visualization of lymphatic vessels and sentinel lymph nodes. Prostate imaging benefits from improved tumor delineation, and vascular applications leverage PAI to assess oxygen saturation and vascular remodeling. In gastrointestinal and hepatic imaging, PAI supports functional assessment and lesion detection with enhanced contrast. Emerging applications in neuro-oncology demonstrate the potential of PAI for intraoperative guidance and brain tumor imaging. Compared to fluorescence imaging, PAI provides deeper penetration and quantifiable contrast. Studies using both approved and investigational agents, including gold nanorods and targeted dye conjugates, highlight advances in imaging tumor margins. Progress in transcranial PAI and molecular probe design continues to broaden its capabilities. Together, these developments underscore the expanding clinical utility of contrast-enhanced PAI for real-time, functional, and molecular imaging.

Pulsed near-infrared (NIR) lasers can be perceived as light of approximately half their wavelength due to the process of two-photon (2P) absorption. For high intensities of light, single-photon (1P) absorption can still be perceived beyond 700 nm, so that the overall color perception in this region is the result of a mix of 1P and 2P absorption. In this study, color matching experiments were performed with seven laser wavelengths between 730 and 920 nm to investigate the interaction between 1P and 2P absorption and the range of colors that can be created by changing the laser power and repetition frequency of a ns-pulsed laser. We recorded color matches in a range from pure red for shorter wavelengths across shades of purple up to pure blue colors for the longest wavelengths, showing that nonspectral (purple) colors can be created using only one stimulating wavelength in the NIR. Changes in hue could be observed between wavelengths of 850 nm and 920 nm when laser power or repetition frequency were modified, with the highest color shifts occurring between 880 nm and 900 nm.

Full-field optical coherence tomography (FFOCT) has recently regained attention thanks to the development of high-resolution dynamic OCT and cross-talk-free swept source FFOCT. However, the choice of wavelength and axial resolution is often a limiting factor with few existing commercial solutions. Here, we developed a novel method to provide rapid spectral shaping for FFOCT imaging. Combining a supercontinuum laser, a fast controllable acousto-optic tunable filter (AOTF), and a multimode fiber with passive and active mode mixing, we obtained an extremely flexible light source compatible with FFOCT. By tuning the AOTF frequency and integrating the resulting wavelength over one camera exposure time, it becomes possible to build any spectrum of interest in the 575-1000 nm range in time domain FFOCT. Alternatively, the designed source module enables achieving swept source FFOCT at up to 100 kfps at an unprecedented axial resolution of 1.1 .