Label-free prediction of three-dimensional fluorescence images from transmitted-light microscopy.

We present the Light My Cells Database, a large-scale open-access collection comprising 2,574 acquisition sets and 56,984 microscopy 2D images designed to support the development of machine learning models for fluorescence prediction from transmitted light images. The dataset aggregates data from 30 independent studies conducted across 8 national imaging centers and captures a wide diversity of biological samples, imaging modalities, and acquisition systems. Each transmitted light image - recorded in bright-field, phase contrast, or differential interference contrast -is paired with at least one fluorescence image labeling key subcellular structures: nucleus, mitochondria, tubulin, or actin. All images are standardized in OME-TIFF format and annotated with rich metadata following REMBI guidelines. A dedicated preprocessing pipeline ensures dimensional harmonization, best-focus plane selection, and consistent file naming. The database reflects the variability encountered in real-life microscopy experiments, making it suited for training and benchmarking generalizable deep learning models. It is accessible via the BioImage Archive and supports a range of downstream applications, including in silico labeling, segmentation, and cell profiling from label-free imaging.

BackgroundThe transmitted light detectors present on most modern confocal microscopes are an under-utilised tool for the live imaging of plant cells. As the light forming the image in this detector is not passed through a pinhole, out-of-focus light is not removed. It is this extended focus that allows the transmitted light image to provide cellular and organismal context for fluorescence optical sections generated confocally. More importantly, the transmitted light detector provides images that have spatial and temporal registration with the fluorescence images, unlike images taken with a separately-mounted camera.ResultsBecause plants often provide difficulties for taking transmitted light images, with the presence of pigments and air pockets in leaves, this study documents several approaches to improving transmitted light images beginning with ensuring that the light paths through the microscope are correctly aligned (Köhler illumination). Pigmented samples can be imaged in real colour using sequential scanning with red, green and blue lasers. The resulting transmitted light images can be optimised and merged in ImageJ to generate colour images that maintain registration with concurrent fluorescence images. For faster imaging of pigmented samples, transmitted light images can be formed with non-absorbed wavelengths. Transmitted light images of Arabidopsis leaves expressing GFP can be improved by concurrent illumination with green and blue light. If the blue light used for YFP excitation is blocked from the transmitted light detector with a cheap, coloured glass filters, the non-absorbed green light will form an improved transmitted light image. Changes in sample colour can be quantified by transmitted light imaging. This has been documented in red onion epidermal cells where changes in vacuolar pH triggered by the weak base methylamine result in measurable colour changes in the vacuolar anthocyanin.ConclusionsMany plant cells contain visible levels of pigment. The transmitted light detector provides a useful tool for documenting and measuring changes in these pigments while maintaining registration with confocal imaging.

Cytometry of Reaction Rate Constant (CRRC) is a method for studying cell-population heterogeneity using time-lapse fluorescence microscopy, which allows one to follow reaction kinetics in individual cells. The current and only CRRC workflow utilizes a single fluorescence image to manually identify cell contours which are then used to determine fluorescence intensity of individual cells in the entire time-stack of images. This workflow is only reliable if cells maintain their positions during the time-lapse measurements. If the cells move, the original cell contours become unsuitable for evaluating intracellular fluorescence and the CRRC experiment will be inaccurate. The requirement of invariant cell positions during a prolonged imaging is impossible to satisfy for motile cells. Here we report a CRRC workflow developed to be applicable to motile cells. The new workflow combines fluorescence microscopy with transmitted-light microscopy and utilizes a new automated tool for cell identification and tracking. A transmitted-light image is taken right before every fluorescence image to determine cell contours, and cell contours are tracked through the time-stack of transmitted-light images to account for cell movement. Each unique contour is used to determine fluorescence intensity of cells in the associated fluorescence image. Next, time dependencies of the intracellular fluorescence intensities are used to determine each cell’s rate constant and construct a kinetic histogram “number of cells vs rate constant.” The new workflow’s robustness to cell movement was confirmed experimentally by conducting a CRRC study of cross-membrane transport in motile cells. The new workflow makes CRRC applicable to a wide range of cell types and eliminates the influence of cell motility on the accuracy of results. Additionally, the workflow could potentially monitor kinetics of varying biological processes at the single-cell level for sizable cell populations. Although our workflow was designed ad hoc for CRRC, this cell-segmentation/cell-tracking strategy also represents an entry-level, user-friendly option for a variety of biological assays (i.e., migration, proliferation assays, etc.). Importantly, no prior knowledge of informatics (i.e., training a model for deep learning) is required.

Cytometry of Reaction Rate Constant (CRRC) is a method for studying cell-population heterogeneity using time-lapse fluorescence microscopy, which allows one to follow reaction kinetics in individual cells. The current and only CRRC workflow utilizes a single fluorescence image to manually identify cell contours which are then used to determine fluorescence intensity of individual cells in the entire time-stack of images. This workflow is only reliable if cells maintain their positions during the time-lapse measurements. If the cells move, the original cell contours become unsuitable for evaluating intracellular fluorescence and the CRRC experiment will be inaccurate. The requirement of invariant cell positions during a prolonged imaging is impossible to satisfy for motile cells. Here we report a CRRC workflow developed to be applicable to motile cells. The new workflow combines fluorescence microscopy with transmitted-light microscopy and utilizes a new automated tool for cell identification and tracking. A transmitted-light image is taken right before every fluorescence image to determine cell contours, and cell contours are tracked through the time-stack of transmitted-light images to account for cell movement. Each unique contour is used to determine fluorescence intensity of cells in the associated fluorescence image. Next, time dependencies of the intracellular fluorescence intensities are used to determine each cell's rate constant and construct a kinetic histogram "number of cells vs rate constant." The new workflow's robustness to cell movement was confirmed experimentally by conducting a CRRC study of cross-membrane transport in motile cells. The new workflow makes CRRC applicable to a wide range of cell types and eliminates the influence of cell motility on the accuracy of results. Additionally, the workflow could potentially monitor kinetics of varying biological processes at the single-cell level for sizable cell populations. Although our workflow was designed ad hoc for CRRC, this cell-segmentation/cell-tracking strategy also represents an entry-level, user-friendly option for a variety of biological assays (i.e., migration, proliferation assays, etc.). Importantly, no prior knowledge of informatics (i.e., training a model for deep learning) is required.

A comparison was made between conventional transmitted light microscopy and confocallaser scanning microscopy (CLSM) as the source of digital images for the measurement of wood cell dimensions by image analysis. When compared with confocal microscopy, transmitted light microscopy using 20 μm thick, safranin stained sections overestimated wall thickness by up to 50% and underestimated lumen area by up to 4% due to the effects of out-of-focus haze. Confocal microscopy using 20 μm thick, safranin stained sections, was found to produce more accurate images of the wood cells compared to transmitted light microscopy using thick sections. Images obtained by optical sectioning were comparable to the quality that might be obtained by thin sectioning of resin embedded wood. For example bordered pit chambers were easily resolved in single optical sections but were not resolved in transmitted light images. Confocal microscopy can be performed on sections as thick as 120 μm and by acquiring optical sections more than 5 μm below the surface of the section distortion of cells caused by sectioning can be avoided. Image quality declined with depth leading to substantial changes in cell dimensions at depths beyond 10-20 μm and significant errors in cell dimensions beyond 80 μm depth. Image brightness was also found to decline with depth, more rapidly in water than in immersion oil. A comparison of measurements of cell dimensions in water and in immersion oil indicated that wall thickness changes significantly during drying but that other dimensions remain almost the same in dry compared to wet seetions.

Label-free imaging of nuclear membrane for analysis of nuclear import of viral complexes

The present study was designed to examine the distribution of interglobular dentine in human tooth roots. The material comprised 17 teeth, of which 3 were premolars extracted for orthodontic reasons from children 10–12 years of age and the other teeth (4 incisors, 3 canines and 7 molars) were extracted for periodontitis from individuals aged 32–63 years. All teeth were free of caries and cervical dentine defects. Ground sections of the teeth cut longitudinally were stained with basic fuchsin and observed by fluorescence and confocal microscopy as well as transmitted light microscopy. Basic fuchsin stained the dentinal tubules, interglobular dentine and the granular layer of Tomes. These structures appeared intense blue to faint violet with transmitted light microscopy, whereas their staining displayed intense fluorescence with fluorescence microscopy. Therefore, the interglobular dentine could be detected more sensitively with fluorescence and confocal microscopy than with transmitted light microscopy. Typical interglobular dentine was present in coronal dentine in most of the teeth. In the radicular dentin, position and size of the interglobular dentine was different among the teeth examined. Most of the teeth had the interglobular dentine in the cervical part of the roots (type A). Two premolars displayed the interglobular dentine in the coronal half of the root (type B). The types A and B contained large interglobular areas. A small amount of interglobular dentine was restricted to the apical half of the roots of two canines and one molar (type C). In contrast to types A and B which were seen at both labial or buccal and lingual sides of roots, the interglobular dentine of type C was seen only at one side, labial or lingual. Some of the tooth roots did not show any interglobular dentine (type D). Most of the incisors, canines and premolar were types A, B, and C, respectively, and the molars were mixed types A, C, and D. These results suggest that the factors affecting dentinogenesis during root formation are unique for each tooth.

Overview of Live‐Cell Imaging: Requirements and Methods Used

The cell cycle is the main process that regulates cell growth and development. The shape and size of cells and organelles change dynamically during the cell cycle. In this paper, deep neural networks (DNNs) were used to predict cell cycle phases from microscopic images. The fluorescent images of the nucleus were predicted from transmitted light image channels from a U-Net model to reduce fluorescent labeling and prevent phototoxicity. The predicted cell nucleus images, along with the transmitted light cell images and fluorescently labeled mitochondria images, were used to train the convolutional neural network ResNet34 to predict the cell cycle stage. Compared with only fluorescently labeled or transmitted light images, convolutional neural networks provide increased prediction accuracy with additional predicted nucleus images from transmitted light images, resulting in improved classification of cell cycle phases.

Segmentation is required to quantify cellular structures in microscopic images. This typically requires their fluorescent labeling. Convolutional neural networks (CNNs) can detect these structures also in only transmitted light images. This eliminates the need for transgenic or dye fluorescent labeling, frees up imaging channels, reduces phototoxicity and speeds up imaging. However, this approach currently requires optimized experimental conditions and computational specialists. Here, we introduce "aiSEGcell" a user-friendly CNN-based software to segment nuclei and cells in bright field images. We extensively evaluated it for nucleus segmentation in different primary cell types in 2D cultures from different imaging modalities in hand-curated published and novel imaging data sets. We provide this curated ground-truth data with 1.1 million nuclei in 20,000 images. aiSEGcell accurately segments nuclei from even challenging bright field images, very similar to manual segmentation. It retains biologically relevant information, e.g. for demanding quantification of noisy biosensors reporting signaling pathway activity dynamics. aiSEGcell is readily adaptable to new use cases with only 32 images required for retraining. aiSEGcell is accessible through both a command line, and a napari graphical user interface. It is agnostic to computational environments and does not require user expert coding experience.

Simultaneous Correlative Light and Electron Microscopy of Samples in Liquid

Microbial interactions in the rhizosphere contribute to soil health, making understanding these interactions crucial for sustainable agriculture and ecosystem management. Yet it is difficult to understand what we cannot see; among the limitations in rhizosphere imaging are challenges associated with rapidly and noninvasively imaging microbial cells over field depths relevant to plant roots. Here, we present a bimodal imaging technique called complex-field and fluorescence microscopy using the aperture scanning technique (CFAST) that addresses these limitations. CFAST integrates quantitative phase imaging using synthetic aperture imaging based on Kramers–Kronig relations, along with three-dimensional (3D) fluorescence imaging using an engineered point spread function. We showcase CFAST’s practicality and versatility in two ways. First, by harnessing its depth of field of more than 100 μm, we significantly reduce the number of captures required for 3D imaging of plant roots and bacteria in the rhizoplane. This minimizes potential photobleaching and phototoxicity issues. Second, by leveraging CFAST’s phase sensitivity and fluorescence specificity, we track microbial growth, competition, and gene expression at early stages of colony biofilm development. Specifically, we resolve bacterial growth dynamics of mixed populations without the need for genetically labeling environmental isolates. Moreover, we find that gene expression related to phosphorus sensing and antibiotic production varies spatiotemporally within microbial populations that are surface attached and appears distinct from their expression in planktonic cultures. Together, CFAST’s attributes overcome commercial imaging platform limitations and enable insights to be gained into microbial behavioral dynamics in experimental systems of relevance to the rhizosphere.

Convolutional neural networks (CNNs) have shown significant success in image recognition and segmentation. Based on a CNN-like U-Net architecture, such a model can effectively predict subcellular structures from transmitted light (TL) images after learning the relationships between TL images and fluorescent-labeled images. In this paper, we focused on building corresponding models of subcellular mitochondrial structures using the CNN method and compared the prediction results derived from confocal microscopic, Airyscan microscopic, z-stack, and time-series images. With multi-model combined prediction, it is possible to generate integrated images using only TL inputs, which reduces the time required for sample preparation and increases the temporal resolution. This enables visualization, measurement, and understanding of the morphology and dynamics of mitochondria and mitochondrial DNA.

Step-by-step guide to post-acquisition correlation of confocal and FIB/SEM volumes using Amira software.

Cardiac myocytes undergo programmed cell death as a result of ischemia/reperfusion (I/R). One feature of I/R injury is the increased presence of autophagosomes. However, to date it is not known whether macroautophagy functions as a protective pathway, contributes to programmed cell death, or is an irrelevant event during cardiac I/R injury. We employed simulated I/R of cardiac HL-1 cells as an in vitro model of I/R injury to the heart. To assess macroautophagy, we quantified autophagosome generation and degradation (autophagic flux), as determined by steady-state levels of autophagosomes in relation to lysosomal inhibitor-mediated accumulation of autophagosomes. We found that I/R impaired both formation and downstream lysosomal degradation of autophagosomes. Overexpression of Beclin1 enhanced autophagic flux following I/R and significantly reduced activation of pro-apoptotic Bax, whereas RNA interference knockdown of Beclin1 increased Bax activation. Bcl-2 and Bcl-x(L) were protective against I/R injury, and expression of a Beclin1 Bcl-2/-x(L) binding domain mutant resulted in decreased autophagic flux and did not protect against I/R injury. Overexpression of Atg5, a component of the autophagosomal machinery downstream of Beclin1, did not affect cellular injury, whereas expression of a dominant negative mutant of Atg5 increased cellular injury. These results demonstrate that autophagic flux is impaired at the level of both induction and degradation and that enhancing autophagy constitutes a powerful and previously uncharacterized protective mechanism against I/R injury to the heart cell.