Refine
Year of publication
Document Type
- Article (38)
- Preprint (4)
- Working Paper (1)
Language
- English (43)
Has Fulltext
- yes (43)
Is part of the Bibliography
- no (43)
Institute
Correlative dynamic imaging of cellular landmarks, such as nuclei and nucleoli, cell membranes, nuclear envelope and lipid droplets is critical for systems cell biology and drug discovery, but challenging to achieve with molecular labels. Virtual staining of label-free images with deep neural networks is an emerging solution for correlative dynamic imaging. Multiplexed imaging of cellular landmarks from scattered light and subsequent demultiplexing with virtual staining leaves the light spectrum for imaging additional molecular reporters, photomanipulation, or other tasks. Current approaches for virtual staining of landmark organelles are fragile in the presence of nuisance variations in imaging, culture conditions, and cell types. We report training protocols for virtual staining of nuclei and membranes robust to variations in imaging parameters, cell states, and cell types. We describe a flexible and scalable convolutional architecture, UNeXt2, for supervised training and self-supervised pre-training. The strategies we report here enable robust virtual staining of nuclei and cell membranes in multiple cell types, including human cell lines, neuromasts of zebrafish and stem cell (iPSC)-derived neurons, across a range of imaging conditions. We assess the models by comparing the intensity, segmentations, and application-specific measurements obtained from virtually stained and experimentally stained nuclei and cell membranes. The models rescue missing labels, non-uniform expression of labels, and photobleaching. We share three pre-trained models (VSCyto3D, VSNeuromast, and VSCyto2D) and a PyTorch-based pipeline (VisCy) for training, inference, and deployment that leverages current community standards for image data and metadata.
Dynamic imaging of landmark organelles, such as nuclei, cell membrane, nuclear envelope, and lipid droplets enables image-based phenotyping of functional states of cells. Multispectral fluorescent imaging of landmark organelles requires labor-intensive labeling, limits throughput, and compromises cell health. Virtual staining of label-free images with deep neural networks is an emerging solution for this problem. Multiplexed imaging of cellular landmarks from scattered light and subsequent demultiplexing with virtual staining saves the light spectrum for imaging additional molecular reporters, photomanipulation, or other tasks. Published approaches for virtual staining of landmark organelles are fragile in the presence of nuisance variations in imaging, culture conditions, and cell types. This paper reports model training protocols for virtual staining of nuclei and membranes robust to label-free imaging parameters, cell states, and cell types. We developed a flexible and scalable convolutional architecture, named UNeXt2, for supervised training and self-supervised pre-training. The strategies we report here enable robust virtual staining of nuclei and cell membranes in multiple cell types, including neuromasts of zebrafish, across a range of imaging conditions. We assess the models by comparing the intensity, segmentations, and application-specific measurements obtained from virtually stained and experimentally stained nuclei and membranes. The models rescue the missing label, non-uniform expression of labels, and photobleaching. We share three pre-trained models, named VSCyto3D, VSCyto2D, and VSNeuromast, as well as VisCy, a PyTorch-based pipeline for training, inference, and deployment that leverages the modern OME-Zarr format.
Dynamic imaging of landmark organelles, such as nuclei, cell membrane, nuclear envelope, and lipid droplets enables image-based phenotyping of functional states of cells. Multispectral fluorescent imaging of landmark organelles requires labor-intensive labeling, limits throughput, and compromises cell health. Virtual staining of label-free images with deep neural networks is an emerging solution for this problem. Multiplexed imaging of cellular landmarks from scattered light and subsequent demultiplexing with virtual staining saves the light spectrum for imaging additional molecular reporters, photomanipulation, or other tasks. Published approaches for virtual staining of landmark organelles are fragile in the presence of nuisance variations in imaging, culture conditions, and cell types. This paper reports model training protocols for virtual staining of nuclei and membranes robust to cell types, cell states, and imaging parameters. We developed a flexible and scalable convolutional architecture, named UNeXt2, for supervised training and self-supervised pre-training. The strategies we report here enable robust virtual staining of nuclei and cell membranes in multiple cell types, including neuromasts of zebrafish, across a range of imaging conditions. We assess the models by comparing the intensity, segmentations, and application-specific measurements obtained from virtually stained and experimentally stained nuclei and membranes. The models rescue the missing label, non-uniform expression of labels, and photobleaching. We share three pre-trained models, named VSCyto3D, VSCyto2D, and VSNeuromast, as well as VisCy, a PyTorch-based pipeline for training, inference, and deployment that leverages the modern OME-Zarr format.