Biological Imaging

Summary: Single-molecule localization microscopy techniques, such as (direct) stochastic optical reconstruction microscopy ((d)STORM) and photo-activated localization microscopy (PALM) enable the visualization of subcellular molecular organization beyond the diffraction limit of conventional light microscopy. Not only is data acquisition rather slow, but the downstream analysis of localization datasets often remains computationally challenging and time-consuming. Consequently, the complexity and duration of data processing often limit experiments to the acquisition and analysis of only small numbers of cells or regions of interest, thereby restricting the statistical power and biological reliability of SMLM studies. To address this limitation, we developed an open-source Python-based package for automated, high-throughput post-processing and quantitative analysis of SMLM localization data, enabling efficient and straightforward handling of extensive datasets with minimal manual intervention. Availability and implementation: dSTORMQuant (source code and documentation) are freely available on GitHub at https://github.com/BCMM-Bielefeld-University/dSTORMQuant under GPL v3 license.

Imaging Mass Cytometry (IMC) combines metal-tagged antibody labelling with laser ablation mass spectrometry to generate highly multiplexed spatial images of tissue sections. However, the area that can be acquired within a single region of interest (ROI) is limited by hardware and software constraints, requiring large tissues to be imaged as multiple tiled ROIs. Reconstructing these ROIs into whole-slide images requires additional processing, while the proprietary .mcd file format can hinder integration with standard bioimage analysis workflows. Here, we present MCD Stitcher, an open-source Python package for converting .mcd files into OME-TIFF images with automated whole-slide stitching. The tool supports rectangular and polygonal ROIs, accommodates variable pixel sizes between ROIs, and uses memory-aware chunked reading during data ingestion to process large datasets on standard workstations. The generated OME-TIFF outputs preserve spatial, channel, and acquisition metadata for downstream analysis in tools such as QuPath, napari, and ImageJ/Fiji. MCD Stitcher provides a reproducible workflow for converting raw IMC data into interoperable image formats, enabling whole-slide spatial analysis without reliance on vendor-specific software.

Modern fluorescence and multiphoton microscopy workflows operate within a heterogeneous ecosystem of file formats, partially overlapping metadata standards, and reader-specific conventions. In practice, this frequently leads to silent axis misinterpretations, loss or corruption of physical voxel size information, and laboratory-specific glue code that is fragile, poorly documented, and difficult to reproduce. OMIO, short for Open Microscopy Image I/O, addresses these issues by providing a lightweight, policy-driven image I/O layer for Python that enforces a canonical, OME-compatible data representation at the API boundary. The central contribution of OMIO is the explicit separation of low-level format access from semantic normalization. Existing reader libraries are used as interchangeable backends for extracting pixel data and available metadata, while OMIO enforces axis conventions, metadata interpretation, and fallback decisions in a centralized and auditable policy layer. This design allows heterogeneous microscopy inputs to be converted into a stable representation without propagating backend-specific assumptions into downstream analysis code. The core design principles of OMIO include canonical axis semantics (TZCYX), robust metadata normalization with explicit and auditable fallbacks, memory-aware operation via optional Zarr-based backends, and workflow-level semantics that extend beyond individual files to folder stacks and BIDS-like project structures. This architecture allows OMIO to orchestrate existing reader libraries into a coherent and reproducible I/O pipeline without replacing or duplicating their functionality. OMIO is implemented as an open-source and community-oriented system in which support for additional file formats and metadata conventions can be added incrementally through modular reader backends. By encouraging the contribution of example datasets, backend extensions, and feature requests, OMIO is designed to evolve alongside emerging acquisition systems while preserving strict semantic guarantees at the interface level. The resulting standardized OME-TIFF outputs are immediately suitable for downstream quantitative analysis and interactive inspection in scientific Python workflows, including workflows based on ImageJ and Napari.

The in vitro scratch wound healing assay is one of the most widely used methods for studying collective cell migration, but converting assay images into reproducible measurements remains a practical bottleneck of manual tracing, local software installation, parameter bookkeeping, and limited visibility into how the wound region was segmented. We present Cytomove, a browser-local software tool for reviewable scratch wound healing assay quantification. Cytomove imports local microscopy images, segments the wound region with an explainable variance-and-threshold pipeline implemented in client-side JavaScript without external image-processing dependencies, displays the segmentation as an inspectable overlay before any number is exported, supports single-image and grouped time-course analysis, and exports wound area, wound area fraction, wound width profile statistics, quality-control labels, and full analysis metadata as CSV, Excel, PNG, and ZIP. All processing runs in the browser or in a desktop package built on the same code; microscopy images never leave the users machine. In a preliminary comparison with the ImageJ/Fiji Wound Healing Size Tool (WHST) across five image sets and 31 paired measurements, Cytomove reproduced wound-area behaviour closely in a clean brightfield comparator sequence (mean absolute percentage error 4.1%, Pearson r = 0.9975) and in a phase-contrast time course approaching closure (median area error 6.6%, r = 0.9984), while surfacing near-closure and real-world acquisition difficulties through overlays and quality-control labels. Informal local testing indicates that typical single-image analysis completes within seconds in a modern browser, with no installation or dependency step. Cytomove lowers installation friction, keeps assay data local, and links every exported number to the segmentation image and parameters that produced it.

Mass spectrometry imaging (MSI) records rich molecular spectra at each pixel, but pathology-oriented interpretation requires visualizations analogous to complementary histopathological stains. We present an expert-aligned framework for constructing multi-view MSI panels. Soft Landmark Contrast Edges (SoLaCE) extracts molecular boundaries directly from high-dimensional spectra. Because standard visualization metrics correlated poorly with rankings from a single expert pathologist, we combine luminance contrast and chromatic diversity with SpecEdge-Dice, a boundary-aware measure of agreement between visualization edges and SoLaCE boundaries. Parametric MiCS+LMC (pMiCS) uses a neural network trained on subsampled data to distill multiple MSI segmentations into a reusable spectral-to-RGB mapping, enabling rapid full-image inference, out-of-sample projection, and more consistent color semantics across aligned images. A concept-based interpretation procedure explains pMiCS outputs through sparse mixtures of spectral concepts. In a blinded benchmark, pMiCS ranked highest among the compared methods. We integrate these components into Virtual Pathology Panels, which use hyperparameter optimization to select high-performing or spatially complementary views. This framework supports future workflows that combine morphology-oriented tissue assessment and molecular analysis within a single MSI acquisition. TeaserVirtual pathology panels transform MSI spectra into complementary views for scalable, interpretable tissue analysis.

Speckle artifacts and isolated foreground pixels are common in fluorescence microscopy and can interfere with segmentation and subsequent quantitative image analysis. Conventional denoising methods often modify image intensities through filtering or smoothing, potentially altering biologically relevant fluorescence signals. We introduce Sparse Pixel Cluster Cleaning (SPC-Clean), a topology-aware method that removes poorly supported foreground pixels through iterative neighborhood analysis of a thresholded mask. SPC-Clean is deterministic, training-free, preserves original fluorescence intensities for practical microscopy workflows.

Alterations of fluorescence properties in retinal pigment epithelium (RPE) cells caused by diseases such as age-related macular degeneration (AMD) highlight the need for detailed analysis of the fluorescent RPE granules at the individual level. Precise segmentation and classification of these granules remain challenging due to their limited visual separability. In this study, we present Classi4RPE, a computational algorithm designed to accurately segment RPE granules and classify them into three categories -- lipofuscin (L), melanolipofuscin (ML), and melanin (M) -- based on fluorescence lifetime imaging data, which provide distinctive contrast. The method is implemented in a custom Python framework and employs seeded watershed segmentation to isolate individual granules. Lipofuscin granules are identified as hyperfluorescent structures with longer lifetimes, while granules with shorter lifetimes are further analyzed based on their spatial lifetime distribution from the center to edge, enabling discrimination of ML from other melanin-rich granules. Our approach achieves high performance, with mean sensitivities of 0.99 for L granules and 0.90 for ML granules, and corresponding specificities of 0.93 and 0.98, respectively, compared to manually annotated ground truth. These results demonstrate the potential of Classi4RPE to surpass human visual limitations and provide a robust tool for quantitative RPE analysis.

Accurate estimation of the contrast transfer function (CTF) of tilt images is a critical first step in cryo electron tomography (cryoET), enabling reliable recovery of high-resolution structural information from thick, heterogeneous specimens. This challenge is especially acute in in situ cryoET, where macromolecules are imaged in their native cellular environment, often at high tilt and through substantial specimen thickness, with correspondingly low signal-to-noise ratios. Although CTF parameters can be later refined using reference-based approaches, accurate initial estimates are critical for downstream processing and the interpretability of tomographic reconstructions, yet they remain difficult to automate. Here, we present Quinoa, a software package designed to address these challenges. Quinoa first validates the tilt geometry and assesses data quality to generate robust initial estimates of defocus and phase shift. These estimates are then refined through optimization of a single global model, enabling precise fitting of the per-image defoci, tilt-dependent astigmatisms, time-dependent phase shifts, the specimen orientation (rotation, tilt and pitch) and the specimen thickness. Notably, and as a key distinguishing feature of this approach is that Quinoa fits equiphase-binned polar power spectra. This substantially reduces the computational cost of optimization without sacrificing accuracy, enabling more progressive and exhaustive refinement passes that further improve robustness. We validated Quinoa using both simulated and experimental data and benchmarked its performance against Warp, Ctfplotter, CTFMeasure, and AreTomo. Our results show that Quinoa is the most robust approach across all simulated cases, maintaining high accuracy even in the simultaneous presence of severe astigmatism, high specimen inclination and variable phase shift. Integrated recovery mechanisms further allow Quinoa to adapt automatically to a wide range of pixel sizes, defoci, astigmatisms and specimen thicknesses. Despite fitting a more complex and dynamic model, Quinoa remains extremely efficient due to extensive GPU acceleration, making it well suited for real-time monitoring during data collection as well as high-throughput offline batch processing. By improving automated CTF estimation in challenging tomographic data, Quinoa supports more accurate structural analysis of cells and tissues in situ.

Microscopy datasets are often spatially sparse, wherein relevant structures occupy only a small fraction of the total field of view (FOV), leaving large regions of background devoid of signal. This inherent inefficiency creates file sizes that are larger than needed, which increases the time needed for computational image data analysis and processing, and means unnecessarily large data volumes. In this work, a classical open-source method for content-aware spatial compaction of microscopy images (CASC) is reported that explicitly removes spatial redundancy by reorganising foreground objects into a new, smaller image. CASC combines adaptive intensity normalisation, statistical thresholding, morphological refinement, and connected-component analysis to isolate foreground structures. These structures are then extracted with contextual padding and repacked into a compact domain using a heuristic shelf-based spatial packing strategy. CASC intentionally destroys the spatial topology of the image but preserves pixel intensities exactly, retaining object-level information. The method achieves demonstrable reductions in image area and background content while maintaining high object-level preservation of biological structures, with a reduction in file size of more than 390-fold shown in real image datasets.

Quantifying vascular architecture in histological whole slide images is needed to study tissue organisation, tumour microenvironment biology, and diseaseassociated vascular remodelling. However, vessel analysis in routine immunohistochemistry remains challenging. Available workflows are often manual, require programming expertise, or lack direct integration with digital pathology platforms. We developed VeSpA (Vessel Spatial Analysis), an open-source pipeline and QuPath extension for automated vessel segmentation and morphometric quantification in CD31-stained whole slide images. VeSpA combines configurable signal extraction, using CMYK Yellow channel extraction by default and optional DAB stain deconvolution for H-DAB images, with automatic or percentile-based thresholding, morphological refinement, contour filtering, and lumen filling to generate vessel masks from standard DAB-stained sections. The QuPath extension includes a graphical interface for selecting annotations, TMA cores, or whole images, configuring segmentation parameters, running the Python backend, and importing vessel objects directly into the QuPath hierarchy. For each detected vessel, VeSpA extracts area, major axis length, minor axis length, eccentricity, centroid, and orientation, while also appending summary measurements to parent annotations and TMA cores. Validation against independent pathologist annotations showed that VeSpA achieved segmentation performance close to inter-rater agreement and outperformed yellow channel prompt-based SAM and zero-shot YOLOv8-seg on overlap-based metrics in the tested dataset. VeSpA integrates vessel segmentation, morphometric feature extraction, and QuPath-based visualisation into a single reproducible workflow for vascular quantification in computational pathology and spatial analysis of histological tissue architecture.

Images in the electron microscope are formed by electron scattering and focusing. The spherical geometry of these processes gives rise to two coherent, conjugate spherical wave fronts, known as Ewald spheres. These spheres are associated with the two halves of the contrast transfer function (CTF), and their widths are determined by the focal gradient through the specimen. To properly correct for the CTF, each half of the CTF must be applied to an image individually and integrated into the reconstruction into the corresponding Ewald sphere. Theory indicates that this dual Ewald sphere reconstruction method should recover the maximal amount of information possible. This method was compared to the other reconstruction methods commonly used: the projection approximation (ignoring the Ewald sphere), the simple insertion and the single sideband methods. In simulated reconstructions the dual Ewald sphere method recovered the most information when the correct half of the CTF is matched to the corresponding Ewald sphere. If the wrong half is matched, the result worse than the projection approximation method. Examining reconstructions from real data indicated that the dual Ewald sphere method performs at least as well as the simple insertion method, but not as good as in simulations. The likely reason is the two-fold ambiguity in the assigned orientations of the particle images, which remains an issue to pursue in further studies. In conclusion, the dual Ewald sphere reconstruction method may offer the best way to calculate very high resolution reconstructions when the micrograph quality warrants it. HighlightsO_LIThe dual Ewald sphere reconstruction corrects for the two halves of the CTF. C_LIO_LIThe signs of the two halves of the CTF must correspond to the focal gradient. C_LIO_LIDetermining the focal gradient for individual particle images remains unresolved. C_LIO_LIComplex reconstructions indicate any real space phases are artifacts. C_LI

Four-dimensional scanning transmission electron microscopy (4D-STEM) enables the acquisition of diffraction patterns at every probe position in a dense array. For imaging applications this approach offers significant benefits in terms of spatial resolution and contrast enhancement. In this work, we present the development of a synchronous scan generator integrated with SerialEM software to enable automation of complex experimental protocols such as tomography. The proposed hardware functions as an interface between SerialEM, the scan controls of the microscope, a fast annular dark-field detector, and a synchronized trigger for a pixelated detector. Our previous implementation, named SavvyScan, relied on a dedicated computer equipped with a multichannel acquisition and signal-generation cards, as well as a separate microcontroller for synchronization. Here, we report a low-cost implementation based on a Red Pitaya board, utilizing direct programming of its embedded FPGA and Linux server components. We provide detailed instructions for system installation and operation, along with practical guidance for modifying the source code. System performance is validated through oscilloscope measurements and imaging of a replica grating sample. The utility of the approach is further demonstrated by generating a 3D electron tomogram of a cryogenic sample of mitochondria from a tilt series of shadow montage projections.

Quantitative analysis of radiation induced DNA double strand breaks (DSBs) and their repair is essential for understanding and eventually contributing to improving radiation-based cancer therapies. Using live-cell microscopy, the formation and resolution of DSBs over time can be followed in individual cells through tracking of foci formed by accumulation of DSB repair proteins. However, manual analysis of such time-lapse datasets is a tedious time-consuming task that is prone to operator bias, affecting the reproducibility. Here, we present LiFT, an automated image analysis pipeline, specifically designed for robust quantification of DSB kinetics in live-cell imaging experiments. To quantify DSB kinetics, our pipeline first segments and tracks cell nuclei without requiring a nuclear stain. After correcting for inter-frame motion through image registration, automatic detection and tracking of foci within these nuclei enables direct quantification of the dynamics of individual repair events. Multiple algorithmic options were implemented for each step of the pipeline, ensuring more general applicability to potentially different imaging setups and applications. We evaluated the pipeline using PLC/PRF/5 cells and demonstrated its generalizability on U2OS-SSTR2 cells. Our results show that LiFT enables reproducible and scalable quantification of DSB dynamics, providing a broadly applicable framework to analyse live-cell imaging data in cancer research. To improve the adoption of LiFT, we made it available as an open-source Python package and provided a graphical user interface to select different methods and adjust method related parameters.

Microscopy-based phenotypic profiling relies increasingly on autonomous, unsupervised feature extraction, yet no existing method explicitly separates shape from texture into dedicated and independent latent subspaces by architectural design. Therefore texture, encoding critical biological information such as protein distribution and intracellular organisation, remains inaccessible as an independent feature domain in standard unsupervised approaches. This represents a fundamental limitation that prevents unbiased phenotypic analysis across biological scales. Here we introduce UDIST (Unsupervised Disentanglement of Shape and Texture), a sequential dual variational autoencoder (VAE) framework that tackles this fundamental limitation by explicitly decoupling shape from texture into independent, non-overlapping latent subspaces at the single-object level. By training two VICReg-regularised VAEs on principal-axis-aligned objects, UDIST separates binary shape from continuous texture information into rotation-invariant feature spaces, enabling separate downstream analysis of both domains. We validated UDIST across biological scales, from nuclei and single cells to patient-derived intestinal organoids, using both fluorescence and brightfield imaging, revealing phenotypic differences previously hidden by morphological variation and enabling the independent analysis of shape and texture in downstream analyses including clustering and similarity measurements. UDIST provides a versatile, label-free, and unsupervised tool for multi-scale phenotypic profiling in high-content microscopy and screening.

Modern biological imaging generates large, complex datasets that require scalable and reproducible image analysis methods. Deep learning has demonstrated strong performance on bioimage segmentation tasks, but training custom models has remained inaccessible to many researchers due to requirements for GPU infrastructure, programming expertise, and large annotated training datasets. ZEISS arivis Cloud is a browser-based platform for deep learning model training that addresses these barriers through partial annotation support, AI-assisted labeling with SAM (Segment Anything Model), pretrained model initialization, and automatically configured training pipelines requiring no machine learning expertise. The platform supports two segmentation tasks: semantic segmentation using a U-Net-style architecture with an EfficientNet encoder and PixelShuffle decoder, and instance segmentation based on Mask2Former with a Swin-Tiny backbone. Both pipelines incorporate microscopy-specific adaptations including smooth tiling, multi-channel input support, dataset-specific normalization, and partial-annotation-aware loss functions protected by patents US-20240078681-A1 and US-20250111519-A1. Trained models integrate directly with ZEISS arivis Pro for pipeline-based image analysis, ZEISS arivis Hub for parallel execution across large datasets, and ZEISS ZEN for content-aware guided acquisition. We describe the platform architecture, training methodology, segmentation architectures, reproducibility and versioning mechanisms, and FAIR compliance, and illustrate the complete workflow through two intestinal organoid imaging examples. arivis Cloud is freely accessible to student users; other users access the platform via subscription at https://www.arivis.cloud/.

Current segmentation models are capable of routine identification of biological features in noisy cryogenic electron microscopy (cryoEM) images. However, there are still challenges with complete segmentation of high boundary, thin objects such as bacterial cell envelopes and flagella. Moreover, ultralow-dose cryoEM images pose as an additional challenge to boundary distinctions between the object and background. Here, we present TileBac, a benchmark dataset of ultralow-dose montage tiles of Pantoea sp. YR343 to segment bacterial inner and outer membranes for evaluation of model effectiveness. We show that foundation models outperform convolutional neural networks at continuous bacterial cell envelope segmentation despite having lower performance metrics. We release the TileBac benchmark dataset on Hugging Face for further insights into model architecture development.

SummaryIntegrating spatial multi-omics data requires coordinated preprocessing, cross-modality alignment and feature registration across modalities that differ in file format, coordinate system and spatial resolution. No existing tool addresses this pipeline end-to-end from raw experimental files till aligned data object. We present FOCUS, an open-source Python package that takes raw data from spatial transcriptomics, mass spectrometry imaging, Raman spectroscopy imaging and brightfield or fluorescence microscopy through modality-specific preprocessing, interactive spatial alignment and resolution-matching registration to a unified MuData object, driven by a single configuration file. Its modular, registry-based architecture allows straightforward extension to additional modalities. FOCUS is accessible via a command-line interface, a browser-based GUI and a Python API. Availability and implementationFOCUS is implemented in Python 3.11, with a browser-based GUI built on a Vue.js 3 frontend served by a Flask backend. Source code, documentation and container recipes are available at https://github.com/sifrimlab/FOCUS; a versioned release is archived on Zenodo (10.5281/zenodo.21700038). Outputs use the AnnData and MuData formats and are directly compatible with the scverse ecosystem.

Integrating tissue-level organisation with sub-cellular resolution and molecular information often requires combining multiple microscopy modalities and scales. However, aligning images acquired with different modalities, settings, or instruments remains challenging. Here, we introduce NucleiSky, a microscopy image registration framework that utilises the spatial arrangement of nuclei or other landmarks as an intrinsic biological fingerprint. NucleiSky represents images as constellations of centroids and aligns them using geometric algorithms and spatial consensus scoring. In benchmark datasets, NucleiSky could localise query regions within larger reference images using as few as five nuclei. We show that NucleiSky can locate high-magnification fields of view within low-magnification overview scans, map these alignments to additional channels, support live brightfield-to-fixed registration using synthetic nuclear labels, and guide microscope retargeting. We further show that the same constellation-matching principle can be extended to 3D localisation and to non-nuclear landmarks. These findings establish local landmark geometry as an intrinsic spatial fingerprint that enables localisation and registration across imaging scales, modalities and microscopy platforms. NucleiSky is available as an open-source Python package and as notebook-based applications.

Most atomic model refinement methods in cryo-EM fit models to the reconstructed density map and effectively treat Fourier voxels as equally reliable. However, the uncertainty in the estimation of Fourier coefficients is highly anisotropic, primarily due to the common variability in SNR in different frequency shells and the distribution of particle images across viewing directions. First-principles arguments suggest that atomic models should be fitted to particle images rather than volumes; this strategy may be computationally demanding. We show that under certain modeling choices, fitting atomic models to weighted volumes is equivalent to fitting directly to particle images. Furthermore, we argue that various proxies can be used to capture this and other sources of uncertainty and distortions. We propose that the principle can be implemented in most atomic model-fitting software with relative ease, using information readily available in existing pipelines. As a proof of concept, we extracted the necessary information from standard RELION runs and fed it into a modified version of Servalcat in which we implemented a reinterpreted version of the idea.

High-content imaging produces thousands of morphological measurements per cell. Interpreting these measurements requires normalization to remove plate effects, statistical tests selected on the basis of data distribution, and control over false discoveries across many features tested at once. MorphoStat is an open-source Python pipeline that applies this sequence of steps automatically. Given a CSV file from CellProfiler or a compatible imaging platform, it removes low-quality wells, normalizes each plate against DMSO controls using a MAD-scaled z-score, routes each feature to a parametric or nonparametric test based on a distributional check, applies Benjamini-Hochberg correction, and writes out results and publication-ready figures. On the BBBC021 benchmark (MCF-7 breast-cancer cells, 632 wells, 473 features), MorphoStat recovered 12 of 13 known mechanism-of-action classes in principal component space, confirming that the normalization and statistical routing work as intended. The tool is available at https://github.com/Almunthir334/morphostat (DOI: 10.5281/zenodo.20354069) under the MIT license.