This PDF file contains the front matter associated with SPIE Proceedings Volume 12848, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Microscopists can minimize the crosstalk between multiple fluorescent channels by using maximally distinctive spectrally sensitive engineered point spread functions (PSFs). We introduce Chernoff Information as a figure of merit for creating such distinctive PSFs for single molecule localization microscopy and the average probability of error in trying to distinguish between PSFs as an evaluation metric for the distinctiveness of multi-color PSFs. This approach can be extended for an arbitrary combination of fluorophore emission spectra and a preferred number of phase-manipulation segments of a phase mask, even at low signal-to-noise ratios.

Hyperspectral imaging combines the characteristics of computer vision and point spectroscopy by obtaining an image with both spatial and spectral information. Therefore, in combination with microscopy, it can increase material discrimination possibilities with respect to regular microscopy imaging. We explore this increased discrimination potential to assess exposure to particle contamination, since workplace exposure to specific particle materials poses well-known health hazards. In this respect, we are focusing on discriminating more health relevant particles such as silica in the respirable size fraction. For this purpose, a particle sampling protocol has been proposed and hyperspectral imaging in combination with transmission microscopy is used for particle material identification. We use a Snapscan visual near-infrared (VNIR) camera providing high spectral and spatial resolution in the 460-900 nm range, 150 spectral bands and up to 7 Mpixels of spatial resolution and high acquisition speed. The hyperspectral microscopy system has been tested for discrimination of fifteen different particle materials, such as silica, coal, dolomite, barite, or rutile, among others. The combined analysis of spatial and spectral information shows potential to accurately discriminate the 15 tested particle materials so far by means of a random forest classifier. In addition, a band relevance analysis is performed showing that only a few specific bands are needed to provide accurate discrimination of the tested materials. The hyperspectral hardware and method presented could lead to a faster exposure assessment than traditional techniques used for occupational exposure estimation.

Blood flow dynamics in microvascular networks are intimately related to the health of tissues and organs. While numerous imaging modalities and techniques have been developed to assess blood flow dynamics for various applications, their utilization has been hampered by limited imaging speed and indirect quantification of blood flow dynamics. Here, we demonstrate direct blood cell flow imaging (DBFI) that provides visualization of individual motions of blood cells over a field of 0.71 mm x 1.42 mm with a time resolution of 0.69 ms (1450 frames per second) without using any exogenous agents. DBFI enables precise dynamic analysis of blood cell flow velocities and fluxes in various vessels over a large field, from capillaries to arteries and veins, with unprecedented time resolution. Several exemplary applications of DBFI illustrate the potential of this new imaging technology.

Recent advancements in sources and detectors operating in the NIR-II wavelengths have driven the emergence of NIR-II intrinsic microscopy. These significant technological strides were necessary because longer wavelengths are known to experience reduced scattering and absorption in biological tissue. Leveraging this optical advantage, the application of the NIR-II spectral domain in microscopy holds the potential to improve the depth of imaging and preserve coherence depth. In this study, we showcase the integration of phase-contrast imaging into a NIR-II reflectance confocal microscope for cortical imaging. By capturing images of cortical cell bodies at depths of up to 800 μm, we demonstrate that the implementation of phase contrast provides clear delineation of cortical cell edges, including myelinated axons, blood vessels, and cortical cell bodies. Additionally, we devised a computational method to enhance dynamic components and generate a digitized vascular network from the acquired images.

Quantitative oblique back illumination microscopy (qOBM) enables 3D quantitative phase imaging (QPI) and refractive index tomography (i.e., optical diffraction tomography, or ODT) in arbitrarily thick scattering samples. Here I outline recent efforts to provide additional functional information with qOBM.

Supercontinuum (SC) lasers in combination with tunable acousto-optic filters facilitate multiwavelength digital holographic microscopy (DHM) with a broad spectral range. However, to achieve optimized quantitative phase imaging (QPI) in off-axis DHM the coherence properties of the tunable light source and the holographic carrier interference fringe pattern formation have to be considered as they are wavelength dependent. We thus determined experimentally the wavelength dependency of coherence length and off-axis carrier fringe frequency in a Michelson interferometer-based multi-spectral DHM setup with a continuously tunable SC laser and moreover analyzed the impact of both parameters on hologram formation and QPI image quality. The results are found in good agreement with theoretical predictions and provide essential information for a reliable adjustment of SC laser based multispectral off-axis DHM systems. Finally, the application of multispectral-DHM for reduction of coherence induced disturbances in quantitative phase images of living pancreatic tumor cells in illustrated.

Light Sheet Fluorescence Microscopy (LSFM) is an ideal tool for imaging model organisms that are hundreds of microns in size, providing high resolution and fast frame rates. In order to extend LSFM to rapid 3D volumetric imaging, a popular method is to add an electrically tunable lens (ETL) in the detection path of the LSFM. But for larger fields of view and higher NA objectives, the ETL introduces aberrations in the system. Here, we developed an LSFM with adaptive optics and an ETL for rapid focusing. We demonstrate that the system enables imaging over a volume of 499 × 499 × 140 μm³ with a volumetric speed of 4 Hz. We apply the system to image neural activity in the zebrafish larvae to capture rare seizure events.

Light-Sheet Fluorescence Microscopy (LSFM) has demonstrated its effectiveness in imaging many model organisms. However, traditional LSFMs often exhibit low resolution and encounter striping artifacts. Here, we introduce a novel LSFM design to address these challenges of imaging large biological samples. Our system employs a single-objective light-sheet geometry, incorporating Structured Illumination Microscopy (SIM) to allow multi-direction light-sheet illumination. This combination aims to rectify striping artifacts and achieve improved lateral resolutions of better than 200 nm over a 277 × 277 × 500 μm³ volume. Our system is demonstrated through imaging beads, C. elegans and cerebellum organoids.

Paroxysmal arrhythmias caused by medications are challenging to be prospectively identified. Zebrafish have emerged as an ideal model organism for screening small molecule compounds to study cardiac abnormalities, due to their rapid development, optical transparency during early stages, and similarities to the human heart. To overcome the challenges associated with observing cardiac abnormalities in zebrafish, we sought to develop a light-field microscope, a rapid imaging method with high photon efficiency, for the volumetric acquisition in a single snapshot. This method, along with its variations, utilizes a multi-lens array (MLA) to capture angular information. We have customized a light-field system and developed a pipeline to incorporate the MLA into the detection path. A program based on wave optics has also been developed to calculate the point spread function at different depths. The program involves two main steps: calculating the wide-field PSF and applying the MLA effect as a mask. This enables us to simulate the impact of the MLA on the imaging system. The comparison between the simulated and experimental data allows for the determination of MLA position. We aim to capture drug-induced arrhythmias in zebrafish larvae using this method, exploring the contractile dysfunction across the atrium and ventricle during multiple cardiac cycles. This research aims to deepen our understanding of the mechanisms underlying these arrhythmias and their connection to drug effects.

Its subcellular resolution and minimal sample exposure make light-sheet microscopy the ideal tool to study biological specimen during their early development. A light-sheet microscope scans the sample with a plane of light and collects fluorescence with an objective orthogonal to the illumination. However, tightly focused Gaussian light-sheets suffer from a shallow depth of focus and are susceptible to scattering-induced aberrations. Light-sheets created by non-diffractive Airy beams can overcome this to yield isotropic sub-cellular resolution over a ten-fold larger field-of-view. Airy beam light-sheets have a characteristically curved structure and a broad transverse structure with side-lobes. Digital deconvolution of the raw data is thus essential to obtain high-fidelity images. Provided that the scan is along the direction of the detection axis, and all recorded data fits within working memory, a simple and efficient Wiener filter can recover accurate 3D images. However, multi-millimeter sized samples must be scanned with a light-sheet that is diagonal to the sample surface. The diagonal movement prevents the use of standard Wiener filtering. Moreover, the associated data sets can become too large to fit within the working memory of a consumer-grade GPU. This demands slow off-line processing, thus breaking a rapid experimental feedback-loop. Here, we investigate the potential of on-the-fly deconvolution of diagonally-scanned Airy light-sheet microscopy.

Image projection through a multimode fiber (MMF) or scattering media has applications ranging from optogenetics to near eye-displays. It requires developing computer-generated holography algorithm to obtain phase pattern of spatial light modulator. In order to accurately project light, conventional methods measured the transmission matrix (TM) of the imaging system by interference. However, it is sensitive to phase instability, easily caused by thermal drift and mechanical vibration. In this work, we proposed to use the TM retrieved from intensity-only measurements and develop a nonlinear optimization algorithm to obtain the displayed phase patterns. Our method formulates the forward model with the retrieved TM, derives the analytical derivative and adopts a second-order optimization method. We validate the improved quality of the projected intensity image by an experiment setup with a MMF.

Determining the achieved spatial resolution in microsphere-assisted microscopy (MAM) has remained an important and challenging task. While fluorescent-based microscopies use fluorescent nano-beads to determine the point spread function (PSF), such a method is not practical in label-free super-resolution techniques. In some previous studies, objects with certain features have been imaged and those feature sizes have been reported as the resolution; however, determining the resolution in that way is not consistent with the definition of resolution based on point sources. Furthermore, the imaging characteristics in MAM may depend on specimen-specific properties, such as materials and substrate. In this study, we propose a robust and objective technique to address these issues. We fabricate arrays of nano-pillars varying in diameter and periodicity, ranging from 50 to 1000 nm. These pillars consist of a 20 nm Ag layer with a 2 nm Cr adhesion layer, deposited on a silicon substrate. The role of specimen-specific properties can be investigated by considering different materials. Fabrication involves the bilayer PMMA resist lift-off process, ensuring accurate deposition of nanometer-scale features to serve as sub-diffraction-limited targets in optical super-resolution microscopy. A deconvolution method that provides a robust platform for resolution measurement was used. By deconvolving the object profile, obtained from scanning electron microscopy, from the image profile, the PSF of the system is determined. Our proposed technique offers a reliable and standardized approach for quantifying resolution, magnification, and field-of-view in MAM or any other label-free microscopy technique, enabling accurate and objective comparison across different novel microscopy techniques.

Analyses of biomedical images often rely on accurate segmentation of structures of interest. Traditional segmentation methods based on thresholding, watershed, fast marching, and level set perform well in high-contrast images containing structures of similar intensities. However, such methods can under-segment or miss entirely low-intensity objects on noisy backgrounds. Machine learning segmentation methods promise superior performance but require large training datasets of labeled images which are difficult to create, particularly in 3D. Here, we propose an algorithm based on the Local Binary Fitting level set method and its application specifically designed to improve the segmentation accuracy for low-contrast structures even with significant noise levels present. The proposed algorithm, the Normalized Local Binary Fitting level set method, shows promise in enhancing the segmentation of low-contrast structures in biomedical images, addressing the limitations of traditional segmentation methods, and offering an alternative to machine learning approaches that require extensive training datasets.

We propose a new method to engineer point spread function (PSF) for 3D single-particle tracking using mica. The imaging system in this method uses a standard wide-field fluorescence microscope; the only difference from the ordinary is using a mica as a substrate to mount the sample instead of the coverslip. This approach would have advantages over easy-toimplement, wide axial range tracking capability, and multiplexity of tracking. We demonstrate 3D single-particle tracking in a homogeneous solution.