Spatial Transcriptomics Methods: A Guide to Experimental Approaches

Spatial transcriptomics (ST) encompasses a group of techniques that preserve the in situ spatial context of RNA molecules within intact tissues, enabling researchers to localize cell types and their associated gene expression patterns. This guide categorizes ST methods into imaging-based approaches such as MERFISH and seqFISH, and sequencing-based approaches such as Visium and Slide-seq, with a focus on their underlying principles, practical trade-offs, and decision criteria for method selection. The content is intended for students, researchers, analysts, and life-science professionals who need a practical framework for choosing and implementing ST experiments.

The Core Distinction Between Imaging-Based and Sequencing-Based ST

The fundamental difference between the two major ST categories lies in how spatial information is captured and how transcripts are identified. Imaging-based spatial transcriptomics uses fluorescence microscopy to visualize and decode RNA molecules directly in tissue sections, while sequencing-based approaches use spatially barcoded capture surfaces followed by high-throughput sequencing to recover transcript identities and positions.

Imaging-based ST has advanced from low-plex single-molecule fluorescence in situ hybridization to a diverse set of highly multiplexed platforms with recent multimodal and pathology-compatible capabilities. These platforms differ substantially in their chemistry, coding strategies, and imaging approaches, yet their biological interpretation often converges on a few notable computational biology problems. The analytical framework that converts raw fluorescence signals or accompanying in situ sequencing data into molecule-level, cell-level, and tissue-level representations involves preprocessing, registration, restoration, feature detection, barcode decoding, molecule calling, cell segmentation, transcript assignment, probabilistic cell typing, spatial-domain inference, and atlas integration. Optical crowding, tissue thickness, panel bias, and multimodal complexity all increase computational difficulty in imaging-based approaches.

Sequencing-based ST methods, by contrast, rely on capturing mRNA from tissue sections onto surfaces with positional barcodes. The tissue is typically sectioned onto a slide with barcoded capture spots, permeabilized to release RNA, and the captured transcripts are then sequenced. The spatial resolution is determined by the size and spacing of the capture spots or beads. These methods generally offer whole-transcriptome coverage without the need for a predefined gene panel, which is a significant advantage for discovery-oriented experiments.

At a Glance: Method Comparison Table

The following table summarizes key characteristics of representative ST methods across both categories. This comparison is intended to support initial method screening decisions, not to serve as a substitute for consulting current platform documentation and published benchmarking studies.

Core Principles of Imaging-Based Spatial Transcriptomics

Single-Molecule Fluorescence In Situ Hybridization Foundations

Imaging-based ST builds on the foundation of fluorescence in situ hybridization (FISH), a technique that has gained an irreplaceable place in microbiology because of its ability to detect and locate a microorganism or a group of organisms within complex samples. FISH has evolved drastically in recent decades, with advances in signal intensity, imaging acquisitions, automation, method robustness, and versatility. This has resulted in a range of FISH variants that give researchers access to information such as complex population composition, metabolic activity, gene detection and quantification, and subcellular location of genetic elements.

In the context of ST, these FISH variants have been adapted for highly multiplexed detection of many genes simultaneously in the same tissue section. The key innovation is the use of combinatorial labeling and sequential hybridization or imaging rounds to uniquely identify each transcript species.

MERFISH and seqFISH Strategies

Multiplexed error-robust fluorescence in situ hybridization (MERFISH) and sequential fluorescence in situ hybridization (seqFISH) represent two major imaging-based approaches. Both use combinatorial barcoding schemes where each gene is assigned a unique barcode composed of combinations of fluorescent labels across multiple hybridization or imaging rounds.

MERFISH uses error-robust barcodes that allow for detection and correction of errors during decoding. The approach involves encoding each gene with a unique combination of readout probes that bind to specific sequences, followed by sequential rounds of imaging to read out the barcode. The error-robust design helps distinguish true signals from noise and misidentification.

seqFISH similarly uses sequential hybridization rounds but may employ different barcode designs and signal amplification strategies. The choice between these approaches depends on factors including the number of genes to be measured, the tissue type, the desired spatial resolution, and the available imaging infrastructure.

Signal Detection and Decoding Workflow

The imaging-based ST workflow involves several critical steps. First, tissue sections are prepared and stained with probe panels targeting the genes of interest. Second, sequential rounds of imaging capture fluorescence signals from the hybridized probes. Third, computational methods register images across rounds, detect individual molecules, and decode the barcodes to assign transcripts to specific genes.

The preprocessing and quality control of imaging-based ST data are essential because they directly affect downstream analysis results. Evaluating segmentation quality and identifying transcript assignment errors are critical components of this workflow. Standardized metrics for assessment and reproducibility of imaging-based ST datasets have been developed, including evaluation of reproducibility, sensitivity, dynamic ranges, signal-to-noise ratio, false discovery rates, cell type annotation, and congruence with single-cell profiling.

Core Principles of Sequencing-Based Spatial Transcriptomics

Capture Surface Technologies

Sequencing-based ST methods use spatially barcoded capture surfaces to capture mRNA from tissue sections. The most widely used commercial platform, Visium, uses slides with capture spots that each contain a unique positional barcode. Tissue sections are placed on the slide, permeabilized to release mRNA, and the captured transcripts are reverse transcribed and sequenced. The resulting data associate each transcript with a spatial barcode, allowing reconstruction of gene expression across the tissue section.

Slide-seq and similar bead-based methods use randomly positioned barcoded beads on a surface, achieving higher spatial resolution than spot-based approaches. The bead positions are determined through a decoding process, and the resolution is limited by bead size and packing density.

Permeabilization and Capture Efficiency

A critical experimental variable in sequencing-based ST is the permeabilization time, which determines how much mRNA is released from the tissue and captured on the surface. Insufficient permeabilization results in low RNA capture, while excessive permeabilization can cause RNA diffusion and loss of spatial resolution. Optimization of permeabilization time is typically performed for each tissue type using a tissue optimization protocol that assesses capture efficiency and spatial signal quality.

Whole-Transcriptome Coverage Advantages

Sequencing-based ST methods provide whole-transcriptome coverage, meaning they do not require a predefined gene panel. This is a substantial advantage for discovery experiments where the genes of interest are not known in advance. The unbiased nature of these methods allows for the identification of novel cell types, states, and spatial patterns that might be missed by panel-based approaches.

Practical Workflow for ST Experiment Design

Step 1: Define the Biological Question and Required Resolution

The first decision point is whether the biological question requires single-cell or subcellular resolution, or whether spot-level or bead-level resolution is sufficient. Questions about cell-cell interactions, niche patterning, and subcellular RNA localization generally require imaging-based methods with high resolution. Questions about tissue-level gene expression patterns, tumor microenvironment composition, and regional differences can often be addressed with sequencing-based methods.

Step 2: Determine Whether a Gene Panel Is Acceptable

If the experiment requires measuring a predefined set of genes, imaging-based methods offer the advantage of high sensitivity and specificity for those genes. If the experiment is exploratory and requires unbiased transcriptome coverage, sequencing-based methods are more appropriate. Some imaging platforms now offer larger panels, but the trade-off between panel size and sensitivity should be evaluated.

Step 3: Assess Tissue Type and Sample Compatibility

Different tissue types present different challenges for ST. Dense tissues with high cell density may suffer from optical crowding in imaging-based methods, which reduces the accuracy of transcript-to-cell assignment. Fatty or fibrous tissues may be difficult to section and permeabilize for sequencing-based methods. Fresh-frozen tissue is generally preferred for both approaches, but some platforms support formalin-fixed paraffin-embedded tissue.

Step 4: Evaluate Computational Resources and Expertise

Both ST categories require substantial computational resources and bioinformatics expertise. Imaging-based methods require image processing, cell segmentation, and barcode decoding. Sequencing-based methods require read alignment, spot or bead-level quantification, and spatial analysis. The availability of computational infrastructure and personnel should be considered in method selection.

Step 5: Consider Integration with Single-Cell RNA-Seq Data

Many ST analysis workflows integrate ST data with single-cell RNA sequencing (scRNA-seq) data to perform cell-type deconvolution and annotation. This integration is a well-established approach that enhances the interpretation of ST data by leveraging the high-resolution cell-type information from scRNA-seq. The choice of deconvolution method and the quality of the scRNA-seq reference data are important factors in the success of this integration.

Options and Trade-Offs in Method Selection

Resolution Versus Coverage Trade-Off

The most fundamental trade-off in ST is between spatial resolution and transcriptomic coverage. Imaging-based methods can achieve subcellular resolution but are limited to measuring a predefined panel of genes. Sequencing-based methods provide whole-transcriptome coverage but at lower spatial resolution. Recent developments in both categories are narrowing this gap, but the trade-off remains a central consideration.

Sensitivity and Dynamic Range

Imaging-based methods generally have high sensitivity for the genes included in the panel, with the ability to detect individual RNA molecules. However, optical crowding in dense tissue can reduce sensitivity because overlapping signals are difficult to resolve. Sequencing-based methods have variable capture efficiency depending on tissue type and permeabilization, and the dynamic range may be limited by sequencing depth.

Reproducibility Across Sites and Platforms

Reproducibility is a critical concern in ST experiments, particularly for multi-site studies. Standardized operating procedures and quality metrics have been developed to evaluate reproducibility across sites and platforms. These include assessments of sensitivity, dynamic range, signal-to-noise ratio, false discovery rates, and congruence with single-cell profiling. Researchers planning multi-site studies should implement these standardized procedures to ensure comparability of results.

Cost and Throughput Considerations

Imaging-based methods require specialized microscopy equipment and may have higher per-sample costs for large panels. Sequencing-based methods require sequencing infrastructure and have costs associated with library preparation and sequencing depth. Throughput considerations include the number of samples that can be processed in parallel and the time required for imaging or sequencing.

Observations and Measurements in ST Experiments

Key Quality Metrics to Track

Several quality metrics should be tracked throughout an ST experiment. For imaging-based methods, these include the number of detected transcripts per cell, the fraction of transcripts assigned to cells, the signal-to-noise ratio, and the false discovery rate. For sequencing-based methods, these include the number of unique molecular identifiers per spot or bead, the fraction of reads mapped to genes, and the spatial autocorrelation of gene expression.

Cell Segmentation Quality Assessment

Cell segmentation is a critical step in imaging-based ST that directly affects downstream analysis. Poor segmentation can lead to transcript misassignment and inaccurate cell-type identification. Quality assessment of segmentation involves comparing segmentation boundaries with histological features and evaluating the consistency of transcript counts within segmented cells. Open-source software tools have been developed to evaluate segmentation quality and identify transcript assignment errors.

Transcript-to-Cell Assignment Accuracy

The accuracy of assigning transcripts to their cell of origin underlies most downstream analysis in imaging-based ST. There is a method-independent ceiling on how accurately this assignment can be recovered, and this ceiling falls below 0.9 in dense tissue. Published methods leave additional accuracy unrecovered beneath this ceiling. Researchers should be aware of these limits when interpreting spatial co-expression claims and other analyses that depend on accurate transcript assignment.

Records and Documentation for ST Experiments

Experimental Metadata Requirements

Comprehensive metadata documentation is essential for ST experiments. This includes tissue source, fixation method, section thickness, probe panel or capture platform version, permeabilization time, imaging parameters or sequencing depth, and software versions for all analysis steps. This metadata supports reproducibility and enables comparison across experiments and sites.

Data Storage and Sharing Considerations

ST datasets are large, particularly imaging-based datasets that include raw image files. Data storage requirements should be planned in advance, and data sharing policies should be considered. The NIH Genomic Data Sharing Policy provides guidance on data sharing expectations for NIH-funded research. FAIR Guiding Principles provide a framework for making data findable, accessible, interoperable, and reusable.

Analysis Pipeline Documentation

Documentation of the analysis pipeline is critical for reproducibility. This includes the specific software tools, parameters, and versions used at each step. Version control and containerization can help ensure that analyses can be reproduced exactly. The EMBL-EBI Training resources provide guidance on bioinformatics best practices, and NCBI Data Resources offer repositories for depositing and accessing genomic data.

Common Failure Patterns in ST Experiments

Low RNA Capture or Detection

Low RNA capture in sequencing-based methods is often caused by insufficient permeabilization, poor tissue quality, or suboptimal capture surface conditions. In imaging-based methods, low detection may result from probe design issues, insufficient hybridization, or suboptimal imaging conditions. Troubleshooting should begin with a review of positive control genes and comparison with expected expression patterns.

Optical Crowding and Signal Overlap

In imaging-based ST, optical crowding occurs when the density of transcripts is too high for individual molecules to be resolved. This is particularly problematic in dense tissues and can lead to underestimation of transcript counts and misassignment of transcripts to cells. Strategies to mitigate optical crowding include using thinner tissue sections, optimizing probe panels to reduce the number of highly expressed genes, and using computational methods to resolve overlapping signals.

Batch Effects and Technical Variation

Batch effects can arise from differences in sample preparation, reagent lots, imaging sessions, or sequencing runs. These effects can confound biological comparisons if not properly controlled. Experimental designs should include batch balancing and appropriate statistical methods to account for technical variation.

Segmentation Errors

Cell segmentation errors are a common source of artifacts in imaging-based ST. Over-segmentation splits single cells into multiple segments, while under-segmentation merges multiple cells into one segment. Both types of errors distort cell-type identification and downstream analyses. Quality control metrics that evaluate segmentation accuracy should be applied before proceeding with downstream analysis.

Limitations and Interpretation Boundaries

Resolution Limits and Their Consequences

The spatial resolution of ST methods imposes limits on the biological questions that can be addressed. Sequencing-based methods with spot sizes of tens to hundreds of microns cannot resolve individual cells in dense tissues. Imaging-based methods can resolve individual molecules but may not accurately assign transcripts to cells in dense tissue due to the recoverability ceiling described earlier.

Panel Bias in Imaging-Based Methods

Imaging-based ST methods are limited to measuring genes included in the panel. This panel bias means that genes not included in the panel cannot be detected, and the selection of genes for the panel can influence the biological conclusions. Panel design should be informed by the biological question and existing knowledge of the tissue or disease of interest.

Computational Complexity and Interpretability

The computational methods used to analyze ST data are complex and can be difficult to interpret. Deep-learning-based models for spatial transcriptome data analysis offer promising avenues for understanding disease mechanisms and expediting drug discovery, but there is an ongoing need to enhance these models for increased biological relevance. Researchers should be cautious about overinterpreting results from complex models without appropriate validation.

Integration Challenges with Other Data Types

Integration of ST data with scRNA-seq, spatial proteomics, and histology data presents significant computational challenges. The choice of integration method can substantially affect the results, and different methods may produce conflicting conclusions. Validation of integration results using orthogonal approaches is recommended.

Quality and Welfare Controls in ST Research

Ethical and Regulatory Considerations

ST research involving human tissues must comply with ethical and regulatory requirements for human subjects research. This includes obtaining appropriate informed consent for tissue use and ensuring compliance with data sharing policies. The NIH Genomic Data Sharing Policy outlines expectations for data sharing and privacy protection in genomic research.

Biosafety Considerations

Tissue handling and processing for ST experiments must follow appropriate biosafety protocols. Fresh-frozen tissues may contain infectious agents, and formalin-fixed tissues require appropriate handling to avoid exposure to fixatives. Institutional biosafety guidelines should be followed for all tissue processing steps.

Data Quality Control Standards

Standardized operating procedures for ST data generation and quality control have been developed to enable evaluation of samples across all technical metrics. These procedures include assessments of reproducibility, sensitivity, dynamic range, signal-to-noise ratio, false discovery rates, cell type annotation, and congruence with single-cell profiling. Implementation of these standards supports comparability across experiments and sites.

Professional Escalation Criteria

When to Consult a Bioinformatics Specialist

Researchers should escalate to a bioinformatics specialist when they encounter challenges in data processing, quality control, or interpretation that exceed their local expertise. Specific situations include unexpected patterns in quality metrics, difficulties in cell segmentation, challenges in integrating ST data with other data types, and uncertainty about the appropriateness of statistical methods.

When to Seek Platform Vendor Support

Platform vendor support should be sought when experimental issues are suspected to arise from the platform itself, including capture surface problems, probe panel performance issues, or instrument malfunctions. Vendors can provide troubleshooting guidance and may offer replacement reagents or services.

When to Consider Alternative Methods

If repeated experiments fail to produce interpretable results with a chosen method, researchers should consider whether an alternative ST method might be more appropriate for their biological question and tissue type. The decision to switch methods should be based on a systematic evaluation of the failure modes and the potential advantages of alternative approaches.

Applications Across Research Domains

Cancer Research and Tumor Microenvironment

ST has been applied extensively in cancer research to characterize tumor heterogeneity, lineage plasticity, and immune microenvironment remodeling. Integrated single-cell and spatial transcriptomic profiling has revealed coordinated reprogramming of the tumor microenvironment during disease progression, including the establishment of immunosuppressive niches and the spatial co-localization of specific cell types. These applications demonstrate the potential of ST to identify therapeutic targets and inform treatment strategies.

Dermatologic Research

ST has been applied in dermatologic research to improve understanding of niche patterning and cell-cell interactions within heterogeneous tissues that encompass skin homeostasis and disease. The ability to localize cell types and their associated gene expression within intact skin tissue provides insights that are not available from bulk or single-cell approaches.

Veterinary Medicine

ST applications in veterinary medicine are emerging, with research, diagnostics, and treatment strategies being explored. The transfer of ST methods to veterinary species presents both opportunities and challenges, including the need for species-specific probe panels and reference data.

Head and Neck Pathology

ST has been applied to head and neck pathology to visualize gene expression within native tissue architecture, enabling insights into cellular heterogeneity, tumor microenvironment composition, and molecular pathways driving disease progression. Artificial intelligence-driven workflows have demonstrated significant applications in this context, including the identification of driver genes and new immunohistochemical biomarkers.

Computational Analysis Approaches

Preprocessing and Quality Control

Preprocessing of ST data involves multiple steps that vary by method category. For imaging-based methods, preprocessing includes image registration, background subtraction, molecule detection, and barcode decoding. For sequencing-based methods, preprocessing includes read alignment, barcode assignment, and unique molecular identifier counting. Quality control metrics should be applied at each step to identify and address technical artifacts.

Cell-Type Deconvolution and Annotation

Cell-type deconvolution methods for ST use reference data, typically from scRNA-seq, to estimate the cell-type composition of each spatial location. These methods are essential for interpreting ST data in tissues with mixed cell populations. The choice of deconvolution method and the quality of the reference data are critical factors in the accuracy of cell-type estimates. Advances and challenges in cell type annotation for spatial transcriptomics continue to be an active area of method development.

Spatial Domain Inference

Spatial domain inference methods identify regions of the tissue with distinct gene expression patterns, effectively segmenting the tissue into functional domains. These methods use spatial information to identify contiguous regions with similar molecular profiles. The resulting spatial domains can be correlated with histological features and used to characterize tissue organization.

Integration with Histology and Pathology

Computer vision-based artificial intelligence approaches are opening new avenues for ST analysis by modeling complex histological patterns and linking morphology to molecular states. These approaches can predict ST directly from histology images, enabling virtual sequencing that reduces costs and integrates morphological insights from pathology with molecular biomarkers. Computer vision techniques can also reconstruct pixel-aligned 3D tissue models, advancing 3D spatial omics analytics.

Three-Dimensional Spatial Transcriptomics

Current Approaches to 3D Reconstruction

Most ST methods analyze two-dimensional tissue sections, but the inherent three-dimensional nature of tissues means that 2D analysis provides an incomplete picture. Computer vision techniques can reconstruct pixel-aligned 3D tissue models from serial sections, overcoming the technical barriers of 2D acquisition and advancing 3D spatial omics analytics. These approaches require careful registration of serial sections and computational methods to align and integrate data across sections.

Challenges in 3D ST

The reconstruction of 3D ST data presents significant challenges, including section-to-section registration, handling of tissue distortion during sectioning, and the computational resources required for 3D data processing and visualization. The reliance on 2D analysis of inherently 3D tissues is a key limitation of current ST methods, and 3D approaches are an active area of development.

Applications of 3D ST

3D ST has potential applications in understanding tissue architecture, developmental processes, and disease progression in their full spatial context. The ability to visualize gene expression patterns in 3D can reveal organizational principles that are not apparent from individual 2D sections.

Frequently Asked Questions

What is the main difference between imaging-based and sequencing-based spatial transcriptomics?

Imaging-based spatial transcriptomics uses fluorescence microscopy to detect and decode RNA molecules directly in tissue sections, providing subcellular resolution but requiring a predefined gene panel. Sequencing-based spatial transcriptomics uses spatially barcoded capture surfaces followed by high-throughput sequencing, providing whole-transcriptome coverage but at lower spatial resolution. The choice between these approaches depends on whether the biological question requires single-cell resolution with a targeted gene set or unbiased transcriptome coverage with lower resolution.

How do I choose between MERFISH and seqFISH for my experiment?

The choice between MERFISH and seqFISH depends on factors including the number of genes to be measured, the tissue type, the desired spatial resolution, and the available imaging infrastructure. MERFISH uses error-robust barcodes that allow for detection and correction of errors during decoding, while seqFISH uses sequential hybridization rounds with potentially different barcode designs. Consultation of current platform documentation and published benchmarking studies is recommended before making a final decision.

What is the typical spatial resolution of Visium compared to imaging-based methods?

Visium capture spots are typically tens of microns in size, meaning each spot may contain multiple cells. Imaging-based methods can achieve subcellular resolution, detecting individual RNA molecules within cells. The appropriate resolution depends on the biological question, with cell-cell interaction studies generally requiring higher resolution than tissue-level pattern analysis.

How important is cell segmentation in imaging-based spatial transcriptomics?

Cell segmentation is critical because it determines which transcripts are assigned to which cells, and this assignment underlies most downstream analysis. Poor segmentation can lead to transcript misassignment and inaccurate cell-type identification. There is a method-independent ceiling on how accurately transcript-to-cell assignment can be recovered, and this ceiling falls below 0.9 in dense tissue. Quality assessment of segmentation should be a standard part of the analysis workflow.

Can I integrate spatial transcriptomics data with single-cell RNA-seq data?

Yes, integration of ST data with scRNA-seq data is a well-established approach for cell-type deconvolution and annotation. The scRNA-seq data provide high-resolution cell-type information that can be used to interpret the spatial data. The choice of deconvolution method and the quality of the scRNA-seq reference data are important factors in the success of this integration.

What are the main quality metrics I should track in a spatial transcriptomics experiment?

For imaging-based methods, track the number of detected transcripts per cell, the fraction of transcripts assigned to cells, the signal-to-noise ratio, and the false discovery rate. For sequencing-based methods, track the number of unique molecular identifiers per spot or bead, the fraction of reads mapped to genes, and the spatial autocorrelation of gene expression. Standardized operating procedures and quality metrics have been developed to enable evaluation of samples across all technical metrics.

How do I handle batch effects in spatial transcriptomics experiments?

Batch effects can arise from differences in sample preparation, reagent lots, imaging sessions, or sequencing runs. Experimental designs should include batch balancing and appropriate statistical methods to account for technical variation. Documentation of experimental metadata and analysis pipeline details supports the identification and correction of batch effects.

What are the limitations of spatial transcriptomics that I should be aware of?

Key limitations include the resolution versus coverage trade-off, panel bias in imaging-based methods, optical crowding in dense tissues, the recoverability ceiling for transcript-to-cell assignment, and the reliance on 2D analysis of inherently 3D tissues. Computational methods are complex and can be difficult to interpret, and integration with other data types presents significant challenges. Researchers should validate results using orthogonal approaches where possible.

Related Bioinformatics Guides

• Modern Transcriptomics: From Bulk RNA-Seq to Single-Cell and Spatial Resolution
• Spatial Transcriptomics: Mapping the Cellular Atlas
• Spatial Transcriptomics Alignment and Cellular Neighborhood Analysis
• Variant Calling in Whole Exome Sequencing (WES): Principles, Algorithms, and Veterinary Applications
• Basecalling Algorithms for Nanopore Sequencing

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• Critical review on biofilm methods.. Critical reviews in microbiology, 2017.
• Spatial transcriptomics data and analytical methods: An updated perspective.. Drug discovery today, 2024.
• From bulk RNA sequencing to spatial transcriptomics: a comparative review of differential gene expression analysis methods.. Human genomics, 2025.
• Integrated single-cell and spatial transcriptomic profiling decodes lineage plasticity and immune microenvironment remodeling in prostate cancer progression.. Molecular cancer, 2026.
• FISH Variants.. Methods in molecular biology (Clifton, N.J.), 2021.
• Research Techniques Made Simple: Spatial Transcriptomics.. The Journal of investigative dermatology, 2022.
• A comprehensive survey of computer vision methods for spatial transcriptomics.. Briefings in bioinformatics, 2026.
• Spatial transcriptomics and artificial intelligence: a scoping review of emerging applications in head and neck pathology.. Head and neck pathology, 2026.
• Imaging-Based Spatial Transcriptomics: Data Interpretation Methods and Biomedical Applications.. 2026.
• A method-independent recoverability ceiling for transcript-to-cell assignment in imaging-based spatial transcriptomics. 2026.
• Imaging-based spatial transcriptomics: methods, preprocessing, and quality control. 2026.
• Standardized metrics for assessment and reproducibility of imaging-based spatial transcriptomics datasets.. 2026.
• Advancing sequencing-based spatial transcriptomics with a comprehensive benchmarking study. Nature Methods, 2024.
• Integrating microarray-based spatial transcriptomics and single-cell RNA-seq reveals tissue architecture in pancreatic ductal adenocarcinomas. Nature Biotechnology, 2020.
• Cell segmentation in imaging-based spatial transcriptomics. Nature Biotechnology, 2021.
• Applications of Spatial Transcriptomics in Veterinary Medicine: A Scoping Review of Research, Diagnostics, and Treatment Strategies. International Journal of Molecular Sciences, 2025.
• Cell-type deconvolution methods for spatial transcriptomics. Nature Reviews Genetics, 2025.
• Advances and Challenges in Cell Type Annotation for Spatial Transcriptomics. Archives of Computational Methods in Engineering, 2026.
• Spatial transcriptomics in cancer research and potential clinical impact: a narrative review. Journal of Cancer Research and Clinical Oncology, 2024.
• From pixels to cell types: a comprehensive review of computational methods for spatial transcriptomics deconvolution. Genomics and Informatics, 2025.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.