LIVECell-A large-scale dataset for label-free live cell segmentation.
Ontology highlight
ABSTRACT:
SUBMITTER: Edlund C
PROVIDER: S-EPMC8440198 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:The success of training computer-vision models heavily relies on the support of large-scale, real-world images with annotations. Yet such an annotation-ready dataset is difficult to curate in pathology due to the privacy protection and excessive annotation burden. To aid in computational pathology, synthetic data generation, curation, and annotation present a cost-effective means to quickly enable data diversity that is required to boost model performance at different stages. In this study, we introduce a large-scale synthetic pathological image dataset paired with the annotation for nuclei semantic segmentation, termed as Synthetic Nuclei and annOtation Wizard (SNOW). The proposed SNOW is developed via a standardized workflow by applying the off-the-shelf image generator and nuclei annotator. The dataset contains overall 20k image tiles and 1,448,522 annotated nuclei with the CC-BY license. We show that SNOW can be used in both supervised and semi-supervised training scenarios. Extensive results suggest that synthetic-data-trained models are competitive under a variety of model training settings, expanding the scope of better using synthetic images for enhancing downstream data-driven clinical tasks.
| S-EPMC10121551 | biostudies-literature
Project description:Despite recent remarkable advances in microscopic techniques, it still remains very challenging to directly observe the complex structure of cytoplasmic organelles in live cells without a fluorescent label. Here we report label-free and live-cell imaging of mammalian cell, Escherischia coli, and yeast, using interferometric scattering microscopy, which reveals the underlying structures of a variety of cytoplasmic organelles as well as the underside structure of the cells. The contact areas of the cells attached onto a glass substrate, e.g., focal adhesions and filopodia, are clearly discernible. We also found a variety of fringe-like features in the cytoplasmic area, which may reflect the folded structures of cytoplasmic organelles. We thus anticipate that the label-free interferometric scattering microscopy can be used as a powerful tool to shed interferometric light on in vivo structures and dynamics of various intracellular phenomena.
| S-EPMC5914294 | biostudies-literature
Project description:BackgroundBecause of its non-destructive nature, label-free imaging is an important strategy for studying biological processes. However, routine microscopic techniques like phase contrast or DIC suffer from shadow-cast artifacts making automatic segmentation challenging. The aim of this study was to compare the segmentation efficacy of published steps of segmentation work-flow (image reconstruction, foreground segmentation, cell detection (seed-point extraction) and cell (instance) segmentation) on a dataset of the same cells from multiple contrast microscopic modalities.ResultsWe built a collection of routines aimed at image segmentation of viable adherent cells grown on the culture dish acquired by phase contrast, differential interference contrast, Hoffman modulation contrast and quantitative phase imaging, and we performed a comprehensive comparison of available segmentation methods applicable for label-free data. We demonstrated that it is crucial to perform the image reconstruction step, enabling the use of segmentation methods originally not applicable on label-free images. Further we compared foreground segmentation methods (thresholding, feature-extraction, level-set, graph-cut, learning-based), seed-point extraction methods (Laplacian of Gaussians, radial symmetry and distance transform, iterative radial voting, maximally stable extremal region and learning-based) and single cell segmentation methods. We validated suitable set of methods for each microscopy modality and published them online.ConclusionsWe demonstrate that image reconstruction step allows the use of segmentation methods not originally intended for label-free imaging. In addition to the comprehensive comparison of methods, raw and reconstructed annotated data and Matlab codes are provided.
| S-EPMC6599268 | biostudies-literature
Project description:Unlocking and quantifying fundamental biological processes through tissue microscopy requires accurate, in situ segmentation of all cells imaged. Currently, achieving this is complex and requires exogenous fluorescent labels that occupy significant spectral bandwidth, increasing the duration and complexity of imaging experiments while limiting the number of channels remaining to address the study's objectives. We demonstrate that the excitation light reflected during routine confocal microscopy contains sufficient information to achieve accurate, label-free cell segmentation in 2D and 3D. This is achieved using a simple convolutional neural network trained to predict the probability that reflected light pixels belong to either nucleus, cytoskeleton, or background classifications. We demonstrate the approach across diverse lymphoid tissues and provide video tutorials demonstrating deployment in Python and MATLAB or via standalone software for Windows.
| S-EPMC10014308 | biostudies-literature
Project description:In recent years, fluorescent nanodiamond (fND) particles containing nitrogen-vacancy (NV) centers gained recognition as an attractive probe for nanoscale cellular imaging and quantum sensing. For these applications, precise localization of fNDs inside of a living cell is essential. Here we propose such a method by simultaneous detection of the signal from the NV centers and the spectroscopic Raman signal from the cells to visualize the nucleus of living cells. However, we show that the commonly used Raman cell signal from the fingerprint region is not suitable for organelle imaging in this case. Therefore, we develop a method for nucleus visualization exploiting the region-specific shape of C-H stretching mode and further use k-means cluster analysis to chemically distinguish the vicinity of fNDs. Our technique enables, within a single scan, to detect fNDs, distinguish by chemical localization whether they have been internalized into cell and simultaneously visualize cell nucleus without any labeling or cell-fixation. We show for the first time spectral colocalization of unmodified high-pressure high-temperature fND probes with the cell nucleus. Our methodology can be, in principle, extended to any red- and near-infrared-luminescent cell-probes and is fully compatible with quantum sensing measurements in living cells.
| S-EPMC7299945 | biostudies-literature
Project description:Imaging of nucleic acids is important for studying cellular processes such as cell division and apoptosis. A noninvasive label-free technique is attractive. Raman spectroscopy provides rich chemical information based on specific vibrational peaks. However, the signal from spontaneous Raman scattering is weak and long integration times are required, which drastically limits the imaging speed when used for microscopy. Coherent Raman scattering techniques, comprising coherent anti-Stokes Raman scattering (CARS) and stimulated Raman scattering (SRS) microscopy, overcome this problem by enhancing the signal level by up to five orders of magnitude. CARS microscopy suffers from a nonresonant background signal, which distorts Raman spectra and limits sensitivity. This makes CARS imaging of weak transitions in spectrally congested regions challenging. This is especially the case in the fingerprint region, where nucleic acids show characteristic peaks. The recently developed SRS microscopy is free from these limitations; excitation spectra are identical to those of spontaneous Raman and sensitivity is close to shot-noise limited. Herein we demonstrate the use of SRS imaging in the fingerprint region to map the distribution of nucleic acids in addition to proteins and lipids in single salivary gland cells of Drosophila larvae, and in single mammalian cells. This allows the imaging of DNA condensation associated with cell division and opens up possibilities of imaging such processes in vivo.
| S-EPMC3516876 | biostudies-literature
Project description:Dopaminergic neurotransmission has been investigated extensively, yet direct optical probing of dopamine has not been possible in live cells. Here we image intracellular dopamine with sub-micrometer three-dimensional resolution by harnessing its intrinsic mid-ultraviolet (UV) autofluorescence. Two-photon excitation with visible light (540 nm) in conjunction with a non-epifluorescent detection scheme is used to circumvent the UV toxicity and the UV transmission problems. The method is established by imaging dopamine in a dopaminergic cell line and in control cells (glia), and is validated by mass spectrometry. We further show that individual dopamine vesicles/vesicular clusters can be imaged in cultured rat brain slices, thereby providing a direct visualization of the intracellular events preceding dopamine release induced by depolarization or amphetamine exposure. Our technique opens up a previously inaccessible mid-ultraviolet spectral regime (excitation ~270 nm, emission < 320 nm) for label-free imaging of native molecules in live tissue.
| S-EPMC4030794 | biostudies-other
Project description:Label-free live single-cell Raman spectroscopy was used to obtain a chemical fingerprint of colorectal cancer cells including the classification of the SW480 and SW620 cell line model system, derived from primary and secondary tumour cells from the same patient. High-quality Raman spectra were acquired from hundreds of live cells, showing high reproducibility between experiments. Principal component analysis with linear discriminant analysis yielded the best cell classification, with an accuracy of 98.7 ± 0.3% (standard error) when compared with discrimination trees or support vector machines. SW480 showed higher content of the disordered secondary protein structure Amide III band, whereas SW620 showed larger α-helix and β-sheet band content. The SW620 cell line also displayed higher nucleic acid, phosphates, saccharide, and CH2 content. HL60, HT29, HCT116, SW620, and SW480 live single-cell spectra were classified using principal component analysis or linear discriminant analysis with an accuracy of 92.4 ± 0.4% (standard error), showing differences mainly in the β-sheet content, the cytochrome C bands, the CH-stretching regions, the lactate contributions, and the DNA content. The lipids contributions above 2,900 cm-1 and the lactate contributions at 1,785 cm-1 appeared to be dependent on the colorectal adenocarcinoma stage, the advanced stage cell lines showing lower lipid, and higher lactate content. The results demonstrate that these cell lines can be distinguished with high confidence, suggesting that Raman spectroscopy on live cells can distinguish between different disease stages, and could play an important role clinically as a diagnostic tool for cell phenotyping.
| S-EPMC6473482 | biostudies-literature
Project description:Label-free quantitative proteomics has become an increasingly popular tool for profiling global protein abundances. However, one major limitation is the potential performance drift of the LC-MS platform over time, which, in turn, limits its utility for analyzing large-scale sample sets. To address this, we introduce an experimental and data analysis scheme based on a block design with common references within each block for enabling large-scale label-free quantification. In this scheme, a large number of samples (e.g., >100 samples) are analyzed in smaller and more manageable blocks, minimizing instrument drift and variability within individual blocks. Each designated block also contains common reference samples (e.g., controls) for normalization across all blocks. We demonstrated the robustness of this approach by profiling the proteome response of human macrophage THP-1 cells to 11 engineered nanomaterials at two different doses. A total of 116 samples were analyzed in six blocks, yielding an average coverage of 4500 proteins per sample. Following a common reference-based correction, 2537 proteins were quantified with high reproducibility without any imputation of missing values from 116 data sets. The data revealed the consistent quantification of proteins across all six blocks, as illustrated by the highly consistent abundances of house-keeping proteins in all samples and the high levels of correlation among samples from different blocks. The data also demonstrated that label-free quantification is robust and accurate enough to quantify even very subtle abundance changes as well as large fold-changes. Our streamlined workflow is easy to implement and can be readily adapted to other large cohort studies for reproducible label-free proteome quantification.
| S-EPMC7425200 | biostudies-literature
Project description:Scarcity of annotated images hampers the building of automated solution for reliable COVID-19 diagnosis and evaluation from CT. To alleviate the burden of data annotation, we herein present a label-free approach for segmenting COVID-19 lesions in CT via voxel-level anomaly modeling that mines out the relevant knowledge from normal CT lung scans. Our modeling is inspired by the observation that the parts of tracheae and vessels, which lay in the high-intensity range where lesions belong to, exhibit strong patterns. To facilitate the learning of such patterns at a voxel level, we synthesize 'lesions' using a set of simple operations and insert the synthesized 'lesions' into normal CT lung scans to form training pairs, from which we learn a normalcy-recognizing network (NormNet) that recognizes normal tissues and separate them from possible COVID-19 lesions. Our experiments on three different public datasets validate the effectiveness of NormNet, which conspicuously outperforms a variety of unsupervised anomaly detection (UAD) methods.
| S-EPMC8544940 | biostudies-literature