Spatial Transcriptomics: Methods, Platforms, and How To Choose

Spatial transcriptomics now spans two fundamentally different measurement strategies, and the choice between them shapes every downstream result a lab will generate. Sequencing-based platforms capture RNA at fixed positions across a slide, while imaging-based platforms read out transcripts one molecule at a time through cycles of hybridization. Neither approach dominates; each trades resolution, throughput, and gene coverage differently, and matching the method to the biological question is the real starting point for any spatial transcriptomics project.

The core measurement problem in spatial transcriptomics

Spatial transcriptomics answers a question that standard single-cell RNA sequencing (scRNA-seq) cannot: where, within an intact tissue section, a given transcript originates. Dissociating tissue into single cells for sequencing-based transcriptomics destroys the spatial relationships between cells, discarding information about tissue architecture, cell neighborhoods, and microenvironment signaling that is often central to the biological question.

The foundational method that established the field, described by Ståhl and colleagues, placed tissue sections directly onto arrayed reverse-transcription primers carrying unique positional barcodes, preserving two-dimensional positional information while generating genome-wide expression data. That single design decision, tagging RNA molecules with a barcode that encodes their spatial origin before sequencing, underlies every capture-based platform that followed.

Imaging-based methods solve the same problem differently. Instead of barcoding RNA for later sequencing, they detect and localize individual transcripts directly in the tissue using rounds of fluorescence hybridization or in situ sequencing chemistry, producing a spatial map without ever removing the RNA from its tissue context.

Transcriptomics is only one layer of a much broader set of tools for understanding spatial biology, which also spans spatial proteomics, epigenomics, and multiomics approaches built on the same principle of preserving a molecule's tissue context. Spatial methods have also benefited from parallel advances in transcriptomics technology, including long-read sequencing and improved annotation of previously uncharacterized transcripts, that continue to expand what any transcriptomic method, spatial or otherwise, can detect.

Earlier spatially aware methods existed before either measurement strategy matured, most notably laser capture microdissection, which lets a researcher physically excise a region of interest from a slide before extracting and sequencing its RNA. That approach preserves spatial information at the level of a manually selected region rather than at single-cell or single-molecule resolution, and it remains useful for bulk profiling of large, well-defined anatomical structures. Modern sequencing-based and imaging-based platforms were built to push resolution far beyond what manual dissection can achieve, down to individual cells or individual molecules.

Sequencing-based spatial transcriptomics: Capturing the whole transcriptome

Sequencing-based (also called capture-based) spatial transcriptomics platforms transfer RNA from a tissue section onto a solid surface patterned with positionally barcoded oligonucleotides, then sequence the captured library and map each read back to its spatial origin using the barcode. The approach preserves the unbiased, whole-transcriptome character of standard RNA sequencing (RNA-seq) while adding a coordinate to every read.

Sequencing-based spatial platforms now span a resolution range that stretches from multi-cell spots down to the scale of individual DNA nanoballs. Bead-based methods such as Slide-seq arrange DNA-barcoded beads of known position beneath a tissue section, reaching roughly 10-micrometer (µm) spatial resolution with substantially improved capture efficiency in later platform versions, approaching that of droplet-based scRNA-seq. Array-based platforms have since pushed toward gapless, sub-cellular feature grids, while DNA nanoball arrays have reached sub-micrometer resolution across large tissue fields, enabling atlas-scale projects that map transcriptional change across whole organs.

Because these platforms sequence essentially the full transcriptome rather than a preselected gene panel, they remain the default choice for discovery-oriented studies where the relevant genes are not yet known. The trade-off is that most capture spots or beads still contain RNA from more than one cell, so computational deconvolution is often needed to assign transcripts back to individual cell types.

Standard array-based capture historically used spots spanning tens of µm in diameter, each averaging several cells' worth of RNA, before newer continuous-grid designs shrank the effective feature size toward single-cell scale. That resolution gain did not require sacrificing whole-transcriptome coverage, which is the core advantage sequencing-based capture retains over every imaging-based alternative, and it is the property that keeps capture-based methods the default choice whenever the full set of relevant genes is not yet known.

Figure 1: A side-by-side comparison of the sequencing-based and imaging-based spatial transcriptomics workflows. Credit: AI-generated image created using Google Gemini (2026).

Imaging-based spatial transcriptomics: Resolving single molecules

Imaging-based spatial transcriptomics methods trade whole-transcriptome breadth for direct, single-molecule visualization. Multiplexed error-robust fluorescence in situ hybridization (MERFISH) encodes each targeted RNA species with a combinatorial barcode read out across successive rounds of hybridization and imaging, a strategy that let its original demonstration image more than 100 RNA species simultaneously in individual cells with error-correcting barcode design. Related sequential fluorescence in situ hybridization (seqFISH) approaches have since scaled to profiling roughly 10,000 genes in single cells while preserving super-resolved subcellular detail, a substantial jump in gene coverage achieved without sacrificing single-molecule accuracy.

A parallel imaging strategy, in situ sequencing, skips combinatorial barcoding altogether and instead sequences short cDNA barcodes directly within fixed tissue using padlock probes and rolling-circle amplification, reading the resulting fluorescent signal through cycles of sequencing-by-ligation chemistry. Because these methods image molecules in place rather than sequencing bulk libraries, they resolve transcripts to a specific location within, or even inside, a single cell.

Imaging-based spatial transcriptomics platforms generally require probes to be designed against a defined gene list in advance, which limits total gene coverage relative to sequencing-based capture. Panel sizes have grown substantially as encoding schemes and imaging throughput have improved, moving from the low hundreds of genes in early demonstrations to panels an order of magnitude larger today. Discovery of an unanticipated transcript entirely outside the designed panel nonetheless remains effectively impossible with current imaging-based chemistry, which is the central limitation researchers weigh against its resolution advantage. Getting this step right is its own discipline: dedicated probe and panel design work, refining probe sequences and panel composition before a full run, is often what separates a successful imaging-based experiment from a wasted tissue section.

Because imaging-based methods localize every detected molecule to a precise position, they resolve biological questions that spot-based capture cannot, including whether a transcript sits inside the nucleus or cytoplasm, or which of two adjacent cells a signal actually originated from. That subcellular precision comes from repeated rounds of hybridization or sequencing chemistry, which also means imaging-based experiments typically take longer per sample and produce larger raw image datasets than a single sequencing run.

Commercial spatial transcriptomics platforms bring both strategies into routine use

Both measurement strategies now have commercialized, standardized instruments built around them. Single-molecule imaging instruments apply combinatorial barcoding or in situ hybridization chemistry inside a packaged workflow, extending subcellular-resolution imaging to laboratories without in-house optics development, and independent benchmarking across multiple such instruments and tissue types has become an active area of platform evaluation research.

A separate category of region-based digital profiling instruments takes a different approach entirely: rather than resolving every transcript, these platforms use ultraviolet-cleavable, barcoded probes released from operator-selected regions of interest, then quantify the released barcodes by sequencing or direct counting. This sacrifices single-cell or single-molecule resolution in exchange for flexibility, letting a researcher profile pathologist-defined regions such as a tumor margin or an immune infiltrate without committing to whole-slide imaging.

Newer single-molecule commercial instruments have pushed localization precision well below the size of a typical cell and expanded gene panels well beyond what early academic demonstrations supported. That combination of nanometer-scale localization and much larger gene panels illustrates how quickly commercial packaging has closed the gap between academic prototypes and routine laboratory workflows, and it is a major reason single-molecule imaging has moved from specialized labs into standard core-facility offerings.

Sequencing-based spatial transcriptomics has followed a parallel commercialization path, moving toward direct integration with next-generation sequencing infrastructure many labs already operate. Newer whole-transcriptome spatial sequencing platforms pair a large, flexible capture area with single-cell resolution and unbiased gene coverage, lowering the barrier for a sequencing core to add spatial profiling without adopting an entirely separate imaging instrument.

Spatial transcriptomics resolution, coverage, and sensitivity trade-offs

Every spatial transcriptomics platform sits somewhere on a triangle defined by three competing properties: spatial resolution, transcriptome coverage, and detection sensitivity. Sequencing-based platforms maximize coverage, capturing tens of thousands of genes without prior selection, but historically sacrifice resolution and per-transcript sensitivity relative to imaging-based methods.

Table 1: A simplified comparison of resolution and coverage characteristics across representative sequencing-based and imaging-based spatial transcriptomics method families.

Method family	Representative approach	Typical spatial resolution	Gene coverage
Sequencing-based (array capture)	Barcoded spot arrays	Multi-cell spot to near single-cell	Whole transcriptome
Sequencing-based (bead capture)	DNA-barcoded bead arrays	Near single-cell	Whole transcriptome
Sequencing-based (nanoball array)	DNA nanoball-patterned arrays	Sub-cellular	Whole transcriptome
Imaging-based (combinatorial FISH)	Multiplexed error-robust hybridization	Single molecule	Hundreds to thousands of targeted genes
Imaging-based (in situ sequencing)	Padlock probe and rolling-circle amplification	Single molecule, sub-cellular	Hundreds to thousands of targeted genes

This comparison is necessarily simplified. Exact resolution and gene coverage figures vary by instrument generation, tissue type, and library preparation choices, so the table should be read as a directional guide to method families rather than a specification sheet for any single commercial product.

Imaging-based platforms invert that trade-off. Resolution and throughput differences between platform families are not simply a matter of newer being better: a gene panel of a few hundred transcripts imaged at true single-molecule resolution answers a different scientific question than an unbiased, whole-transcriptome capture experiment binned at multi-cell resolution, even when both experiments run on the same tissue block.

Sensitivity, meaning the fraction of true transcripts a platform actually detects and reports, has improved substantially across both families as chemistry has matured, as detailed above for bead-based capture, while imaging-based methods have separately scaled gene panels from roughly 100 targets to several thousand without losing single-molecule accuracy.

Choosing a spatial transcriptomics method for your research question

The practical decision rests on what the experiment needs to answer rather than on which platform is newest. A discovery-stage project asking which genes distinguish two tissue regions, with no strong prior hypothesis, is better served by an unbiased sequencing-based capture platform, even at coarser spatial resolution, because it will not silently miss a relevant transcript outside a predefined panel.

This choice affects more than the first experiment. A platform selected for an initial pilot study often becomes the default for an entire research program, since downstream analysis pipelines, reference datasets, and staff training accumulate around whichever method a lab adopts first. Revisiting that choice mid-program is costly, so it is worth weighing the coverage-versus-resolution trade-off before the first sample is processed, not after a program is already built around one method family.

A hypothesis-driven project asking how a defined set of genes is organized at the subcellular or single-cell level, such as mapping receptor-ligand pairs across a tissue microenvironment, is better served by an imaging-based or single-molecule platform, where resolution and detection accuracy on the targeted genes matter more than total transcriptome breadth. Many labs now run both strategies in sequence: an unbiased capture experiment to nominate candidate genes, followed by a targeted imaging panel built around those candidates for high-resolution validation.

A short evaluation framework helps narrow the decision before committing budget and tissue to a single platform:

• Define whether the question requires whole-transcriptome discovery or validation of a known gene set, since that alone rules out roughly half of available platforms.
• Estimate the resolution the biology actually requires; subcellular precision is not necessary for every question and adds cost and complexity when it is not.
• Check tissue preparation compatibility, since fixation method, tissue thickness, and species can each restrict which platforms are validated for a given sample type.
• Confirm computational and analysis support is available in-house or through a core facility, since deconvolution, cell segmentation, and integration workflows differ substantially between method families.
• Build in controls, QC, and troubleshooting steps from the first pilot run, since catching a failed capture or hybridization step early is far cheaper than discovering it after a full study is already underway.
• Budget for iteration; many spatial transcriptomics projects require a pilot run before a full study design is finalized.

Two further comparisons should factor into platform selection. Understanding how spatial transcriptomics and scRNA-seq complement each other clarifies when a dissociation-based single-cell reference is still the right tool, since the two answer overlapping but distinct questions and are increasingly used together rather than as substitutes. Evaluating resolution, sensitivity, and throughput differences side by side across the full platform landscape, rather than reading individual platform specification sheets in isolation, remains the most reliable way to match a method to a specific tissue, budget, and scientific question.

Building a spatial transcriptomics strategy around the right method

Spatial transcriptomics is not one technology but a family of methods built around two distinct measurement strategies, each with a clear set of strengths and limits. Sequencing-based capture platforms deliver unbiased, whole-transcriptome coverage at spot or bead resolution, while imaging-based and in situ sequencing platforms deliver single-molecule precision across a defined gene panel.

Matching a platform to a research question, rather than defaulting to whichever instrument is locally available, remains the single decision that most affects whether a spatial transcriptomics experiment succeeds. As resolution, sensitivity, and gene coverage continue to converge across both measurement families, the practical differences between sequencing-based and imaging-based spatial transcriptomics will keep shaping which platform is the right choice for a given tissue and question.

For most labs, that decision is worth revisiting as new instrument generations reach the market rather than being made once and left unexamined. A platform choice made two or three years ago may no longer reflect the current resolution, coverage, and cost trade-offs across sequencing-based and imaging-based spatial transcriptomics.

This content includes text that has been created with the assistance of generative AI and has undergone editorial review before publishing. Technology Networks' AI policy can be found here.