Cell Image Analysis Techniques

This paper develops and evaluates Deep Learning models for classifying retinal cell types from cross-sectional images, showing improved performance with extended datasets.

Brainways automates registration, quantification, and network analysis of brain activity from fluorescence images, improving accuracy and efficiency in studying neural circuits.

The paper develops a semi-automated method to segment and model blood cells in zebrafish larvae using X-ray histotomography, enabling quantitative geometric analysis across developmental stages and mutants.

CaliAli improves neuronal tracking in calcium imaging by aligning sessions before extracting features, enhancing detection of weak signals and enabling long-term dynamics analysis.

The paper introduces a method to track neurons across multiple in-vivo imaging sessions without requiring full field-of-view registration, validated on simulated and experimental datasets.

The paper develops an imaging pipeline to quantify 3D cellular proliferation and morphology during mouse facial development, showing their tight correlation.

SKOOTS segments mitochondria in 3D imaging data efficiently, revealing structural changes not detectable with existing methods.

ASTRA is a deep learning tool for automated segmentation of astrocytic calcium signals in two-photon imaging, outperforming existing methods and enabling large-scale analysis.

This paper develops a deep learning classifier to distinguish normal from abnormal cell division in mitosis videos, achieving high accuracy in classification.

The paper introduces u-Unwrap3D to map 3D cell surfaces and analyze interactions between geometry and molecular signals, revealing new dynamics in immune and tumor cell behaviors.

The paper introduces MID, a deep learning method for high-content image analysis that outperforms existing tools in drug toxicity classification and compound identification.

BetaBuddy automates β-cell calcium spiking analysis using deep learning, improving accuracy and efficiency compared to manual methods.

Cell Painting images and profiles for 115,796 compounds in U2OS cells from 12 laboratories; 15.2% of analyzed compounds are phenotypically active, and cross-laboratory replicates cut reference-compound activity by 34%.

The paper develops a weakly supervised U-Net model for precise segmentation of whole brain immunolabeled cell signals, revealing distinct neural activity maps related to social motivations.

The paper introduces a dataset and method for nuclear segmentation and tracking in mouse embryos, identifying Stardist-3D as the most accurate method for this task.

The paper develops a deep learning method to quantify lipophagy in yeast using fluorescence and cryo-soft X-ray microscopy data.

GILEA uses GAN inversion and latent eigenvalue analysis to profile and edit phenomes from biomedical images, demonstrating its effectiveness with SARS-CoV-2 and skin lesion datasets.

RxRx3 provides a large-scale high-content screening dataset combining CRISPR knockouts and small molecule treatments to enable machine learning research and biological discovery.

The paper introduces pAPRica, a framework for efficient analysis of large-scale histopathology images, achieving significant speedup and data reduction.

This paper shows self-supervised vision transformers can accurately decode cellular state heterogeneity from microscopy data without prior training.

This paper develops a deep learning pipeline for automated dendritic spine detection in 3D live cell imaging data, achieving near-human performance without retraining.

AimSeg is a machine learning tool for segmenting axons, inner tongue, and myelin in EM images, improving efficiency and accuracy over manual methods.

Vessel Metrics is a software tool that automates vascular structure analysis in confocal images, improving consistency and efficiency across research groups.

STrack is a tool for tracking bacterial cells in time-lapse microscopy images, showing high consistency across six bacterial strains.

This paper introduces O2-VAE, an orientation-invariant representation learning method for morphologic profiling of cells and organelles, improving clustering and outlier detection.

COMBINe automates detection and classification of neurons and astrocytes in cleared mouse brains using RetinaNet to analyze MADM-based genetic modifications.

The paper uses GANs to generate synthetic cell nuclei and masks for training image segmentation models, demonstrating effectiveness with small datasets.

PIQUE is an ImageJ plugin that automates the quantification of toroidal nuclei in biological images, matching manual counts and identifying micronuclei for CIN analysis.

This paper proposes an RNN-based method for automatic cell-cycle stage classification using time-lapse microscopy images, achieving superior performance over a classifier baseline.

SalienceNet enhances nuclei saliency in low-light microscopy images, improving segmentation accuracy by 30% with Otsu thresholding and 26% with StarDist.