TrueSpot: A robust automated tool for quantifying signal puncta in fluorescent imaging

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 14, 2025

Abstract Characterizing the movement of biomolecules in single cells quantitatively is essential to understanding fundamental biological mechanisms. RNA fluorescent situ hybridization (RNA-FISH) a technique for visualizing fixed using probes. Automated processing resulting images large datasets. Here we demonstrate that our RNA-FISH image tool, TrueSpot, useful automatically detecting locations at molecule resolution. TrueSpot also performs well on with immunofluorescent (IF) and GFP tagged clustered protein targets. Additionally, show 3D spot detection approach substantially outperforms current 2D algorithms.

Language: Английский

---

A Foundation Model for Cell Segmentation

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Nov. 20, 2023

Abstract Cells are a fundamental unit of biological organization, and identifying them in imaging data – cell segmentation is critical task for various cellular experiments. While deep learning methods have led to substantial progress on this problem, most models use specialist that work well specific domains. Methods learned the general notion “what cell” can identify across different domains proven elusive. In work, we present CellSAM, foundation model generalizes diverse data. CellSAM builds top Segment Anything Model (SAM) by developing prompt engineering approach mask generation. We train an object detector, CellFinder, automatically detect cells SAM generate segmentations. show allows single achieve human-level performance segmenting images mammalian (in tissues culture), yeast, bacteria collected modalities. has strong zero-shot be improved with few examples via few-shot learning. also unify bioimaging analysis workflows such as spatial transcriptomics tracking. A deployed version available at https://cellsam.deepcell.org/ .

Language: Английский

---

Piscis: a novel loss estimator of the F1 score enables accurate spot detection in fluorescence microscopy images via deep learning

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 31, 2024

Single-molecule RNA fluorescence in situ hybridization (RNA FISH)-based spatial transcriptomics methods have enabled the accurate quantification of gene expression at single-cell resolution by visualizing transcripts as diffraction-limited spots. While these generally scale to large samples, image analysis remains challenging, often requiring manual parameter tuning. We present Piscis, a fully automatic deep learning algorithm for spot detection trained using novel loss function, SmoothF1 loss, that approximates F1 score directly penalize false positives and negatives but differentiable hence usable training approaches. Piscis was tested on diverse dataset composed 358 manually annotated experimental FISH images representing multiple cell types 240 additional synthetic images. outperforms other state-of-the-art methods, enabling accurate, high-throughput FISH-derived imaging data without need

Language: Английский

---

Automated classification of cellular expression in multiplexed imaging data with Nimbus

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: June 3, 2024

Abstract Multiplexed imaging offers a powerful approach to characterize the spatial topography of tissues in both health and disease. To analyze such data, specific combination markers that are present each cell must be enumerated enable accurate phenotyping, process often relies on unsupervised clustering. We constructed Pan-Multiplex (Pan-M) dataset containing 197 million distinct annotations marker expression across 15 different types. used Pan-M create Nimbus, deep learning model predict positivity from multiplexed image data. Nimbus is pre-trained uses underlying images classify types, tissues, acquired using microscope platforms, without requiring any retraining. demonstrate predictions capture staining patterns full diversity Pan-M. then show how can integrated with downstream clustering algorithms robustly identify subtypes have open-sourced community use at https://github.com/angelolab/Nimbus-Inference .

Language: Английский

---

Guidestar: a spike-in approach to improve RNA detection accuracy in imaging-based spatial transcriptomics

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 10, 2025

Summary Imaging-based spatial transcriptomics technologies, such as MERFISH and seq-FISH, use combinatorial barcoding imaging to simultaneously detect individual RNA molecules from 10s 10,000s of genes. These technologies require the decoding molecules’ location gene identity stacks images. However, beyond using ‘blank’ code-words negative controls, there is a lack ground truth information embedded within assay experimentally measure accuracy sensitivity algorithm. We introduce Guidestar, system spike-in controls integrated FISH assay, that labels subset transcripts with additional probes. probes are imaged separately ‘guide bits’, which serve ground-truth data assess at level molecules. Using Guidestar evaluate an existing method suggested alternative parameter settings increased minimal impact on accuracy. also used dataset train machine-learning based classifier distinguish true false calls, yielding 9% 40% higher F1 scores across cell line tissue samples, respectively.

Language: Английский

---

Artificial intelligence in traditional Chinese medicine: advances in multi-metabolite multi-target interaction modeling

Frontiers in Pharmacology, Journal Year: 2025, Volume and Issue: 16

Published: April 15, 2025

Traditional Chinese Medicine (TCM) utilizes multi-metabolite and multi-target interventions to address complex diseases, providing advantages over single-target therapies. However, the active metabolites, therapeutic targets, especially combination mechanisms remain unclear. The integration of advanced data analysis nonlinear modeling capabilities artificial intelligence (AI) is driving transformation TCM into precision medicine. This review concentrates on application AI in target prediction, including multi-omics techniques, TCM-specialized databases, machine learning (ML), deep (DL), cross-modal fusion strategies. It also critically analyzes persistent challenges such as heterogeneity, limited model interpretability, causal confounding, insufficient robustness validation practical applications. To enhance reliability scalability future research should prioritize continuous optimization algorithms using zero-shot learning, end-to-end architectures, self-supervised contrastive learning.

Language: Английский

---

SpotMAX: a generalist framework for multi-dimensional automatic spot detection and quantification

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 23, 2024

Abstract The analysis of spot-like structures is a widespread task in microscopy-based cell biology. Existing solutions are typically specific to single applications and do not use multi-dimensional information from 5D datasets. Therefore, experimental scientists often resort subjective manual annotation. Here, we present SpotMAX, generalist AI-driven framework for automated spot detection quantification. SpotMAX leverages the full scope datasets with an easy-to-use interface embedded segmentation tracking. outperforms state-of-the-art tools, some cases, even expert human annotators. We applied across diverse questions, ranging meiotic crossover events C. elegans mitochondrial DNA dynamics S. cerevisiae telomere length mouse stem cells, leading new biological insights. With its flexibility integrating AI workflows, anticipate that will become standard microscopy data. Source code: https://github.com/SchmollerLab/SpotMAX

Language: Английский

---

Homebuilt Imaging-Based Spatial Transcriptomics: Tertiary Lymphoid Structures as a Case Example

Methods in molecular biology, Journal Year: 2024, Volume and Issue: unknown, P. 77 - 105

Published: Nov. 11, 2024

Language: Английский

---

Machine learning approaches for spatial omics data analysis in digital pathology: tools and applications in genitourinary oncology

Frontiers in Oncology, Journal Year: 2024, Volume and Issue: 14

Published: Nov. 29, 2024

Recent advances in spatial omics technologies have enabled new approaches for analyzing tissue morphology, cell composition, and biomolecule expression patterns

Language: Английский

---