Cell Image Analysis Techniques

The paper introduces BEN, a method to correct batch effects in deep learning for microscopy images by aligning experimental and deep learning batches, improving model generalization.

This study compares the predictive performance of models trained on fluorescence and brightfield images for drug mechanism of action prediction, finding comparable results.

The paper introduces a deep-learning method called TPS to quantify zonal gene expression patterns in tissues like the liver, showing it can detect differences between strains, ages, and disease states.

The paper introduces nCorrect, an algorithm that improves neuron tracing by reanalyzing existing traces to produce more physiologically-relevant reconstructions using a 4D minimization algorithm and a new similarity metric.

The paper proposes a hybrid deep learning method to exclude non-chromosome objects from metaphase chromosome images, achieving high specificity in segmentation.

H5CellProfiler is an R package that enables user-friendly analysis of large single-cell datasets on a local desktop with efficient memory usage and track labeling.

The paper introduces NuCLear, a method combining in vivo two-photon imaging and deep learning to monitor tissue composition over time in mice brains.

GOAT is a deep learning tool for annotating organoids from microscopy images, showing high performance across diverse organoid systems.

The paper introduces a fisheye transformation for single-cell phenotyping, improving accuracy by weighting nearby cells more heavily than distant ones, and validates it on microscopy and wildlife datasets.

The paper uses electron microscopy data to classify cell types in the neocortex based on perisomatic features, showing that these features are sufficient for accurate classification even in large datasets.

NeuVue is a framework for managing and prioritizing proofreading tasks in connectomic datasets, enabling scalable and coordinated error correction by over forty proofreaders.

The paper introduces ZephIR, a semi-supervised multiple object tracking framework for 2D/3D videos, demonstrating its versatility in tracking biological systems.

This paper develops a machine learning pipeline for detecting bacteria in electron microscopy images of brain tissue, demonstrating its potential for large-scale analysis and quality control.

BOSS segments and tracks oligodendrocytes in 3D images over time with minimal human input, performing similarly to a blinded human observer.

The paper uses geometric deep learning to analyze 3D cell shapes from melanoma cells, identifying morphological signatures and predicting small molecule mechanisms.

This paper introduces SynapseCLR, a contrastive learning method for 3D EM data to extract synapse features from mouse visual cortex, achieving high accuracy in synapse classification and annotation.

The BioImage Model Zoo provides a community standard, infrastructure, and open repository for sharing, testing, and deploying pre-trained deep learning microscopy models across multiple bioimage analysis tools.

The paper explores the use of multiset similarity indices to quantify colocalization between concentrations, suggesting they can provide complementary information to Pearson correlation.

The paper develops a method to quantify 3D concentration gradients and cellular organisation in curved epithelial tissues using image processing and intensity correction.

This study benchmarks deep learning and feature engineering methods for analyzing high-content microscopy images to predict phenotypes and cluster cells without manual annotations.

Pytri is a system using multiple machine learning models to detect biological entities with higher accuracy than manual or OpenCV-based methods.

The paper introduces MorphoGNN, a graph neural network method for embedding single neuron morphology, achieving competitive performance in classification and retrieval tasks.

This paper introduces data-driven microscopy (DDM), a framework that enhances imaging fidelity by combining population-wide characterization with adaptive imaging techniques.

This paper introduces an automated method for quantifying chemotaxis index in C. elegans using liquid-handling robots and a custom image analysis tool.

This paper develops a deep learning method called RELIC to differentiate between normal and pathological kidney exfoliated cells using urine samples with 82% accuracy.

The paper introduces an improved CNMF-based algorithm for calcium imaging data analysis, enhancing cell detection accuracy and signal quality.

The paper introduces tools for automated identification and registration of anatomical landmarks in C. elegans using brightfield images, enabling high-throughput morphometric analysis.

The paper introduces a new volumetric estimation method (TCS) and a tool (VolumestJ) for analyzing brain structure volumes from tissue sections, showing improved accuracy over the classical Cavalieris method.

This paper develops a framework combining CNN and unsupervised learning to map genotype to vacuole morphology in yeast, identifying 4 phenotypic groups with 85% purity and 80% prediction accuracy.

The paper develops a deep learning method with targeted augmentation to track neurons in moving C. elegans, reducing manual annotation needs and enabling analysis of neuronal behavior.