Imaging-Based vs. Sequencing-Based Spatial Transcriptomics: Choosing the Right Platform

• Imaging-based platforms (Xenium®, CosMx®, MERFISH®) directly visualize a targeted gene panel in situ, delivering single-cell or subcellular resolution at the cost of whole-transcriptome coverage.
• Sequencing-based platforms (Visium®, Stereo-seq™) capture and sequence RNA from spatially barcoded spots, giving whole-transcriptome or near-whole-transcriptome coverage at lower spatial resolution.
• Two recent neurodegeneration studies used both approaches in the same project, and neither could have reached its conclusions with just one platform.^1,2
• iSCanGuide™ automates spatial data analysis from both platform types directly, and seamlessly feeds spatial output into iPathwayGuide™ for pathway-level interpretation without a separate export step.

Every spatial transcriptomics project starts with the same question: which platform should be used? The field has settled into two dominant technical approaches with differences in resolution and transcriptome coverage, important considerations that determine what a dataset can and cannot answer well. Here, we define the tradeoffs to consider before selecting a platform for your spatial experiments and explore the latest analytical advancements simplifying spatial data analysis.

Imaging-based spatial transcriptomics platforms label RNA transcripts with probes and then image them directly inside intact cells, so the spatial coordinate of each transcript is captured at the moment of detection rather than reconstructed afterward. The Xenium platform, for example, labels tissue RNA with circularizable DNA probes targeting a defined gene panel, then runs the tissue through successive rounds of fluorescent probe hybridization, imaging, and probe removal to convert each optical signature into a gene identity.¹

Because location is measured directly rather than inferred, imaging-based platforms deliver single-cell or subcellular resolution. That precision makes them well suited to questions about cell-cell contact, receptor-ligand proximity, and the distribution of specific transcripts within a single cell. The limitation lies in panel size: these platforms typically profile a targeted set of hundreds to a few thousand genes rather than the full transcriptome. Common platform examples include Xenium, CosMx, and MERFISH.

Sequencing-based platforms take a different approach. RNA is captured from tissue that sits over a slide printed with spatially barcoded capture spots, then sequenced off-slide and computationally mapped back to its spot of origin. Visium, for instance, arranges 55-micron capture spots in a grid across the tissue section, each barcoded so that every read can be traced back to its spatial location after sequencing.²

Because sequencing captures poly-adenylated RNA broadly rather than a preselected panel, sequencing-based platforms deliver whole-transcriptome or near-whole-transcriptome coverage, which supports unbiased differential expression and pathway-level analysis across a tissue section. The limitation is resolution: standard Visium spots typically span one to ten cells rather than resolving individual cells.² Common platform examples include Visium and Stereo-seq.

Table 1. Spatial transcriptomics platform comparison (resolution ranges drawn from Khan et al., 2025).³

Two recent studies illustrate what each approach contributes when the question is neurodegeneration, and what happens when a project uses both.

Kolling et al.2 used Visium, a sequencing-based platform, to profile the dorsal raphe nucleus in hTau mice, a model of early Alzheimer’s disease. Computational sub-clustering split the region into dorsolateral, centromedial, and ventrolateral subregions, and differential expression analysis identified genes tied to kinase binding, ion channel activity, and serine/threonine kinase regulation. One finding stood out: the long noncoding RNA Map2k3os was enriched specifically in the dorsolateral subregion of hTau mice, tracking with both Tau pathology and serotonergic neuron loss in that same subregion, a pattern the authors then validated with RNAscope® and immunohistochemistry.
In this study, manual methods were used to identify a list of differentially expressed genes which were subsequently run through Advaita’s iPathwayGuide™ to sort them into relevant cellular, biological, and disease pathways.

Wen et al.¹ combined Visium with Xenium, an imaging-based platform, to study Purkinje cell-specific MANF knockout mice in a binge alcohol exposure model. Manual analysis of Visium data identified differentially expressed genes across the Purkinje cell layer, and those DEGs were then run through iPathwayGuide™ for enrichment analysis, surfacing genes tied to protein folding and endoplasmic reticulum stress response. Xenium then profiled a custom 300-gene panel at single-cell resolution within those same tissue sections, confirming the top Visium hits (including Hspa5 and Sdf2l1) and adding transcript-level detail that Visium’s spot resolution could not resolve on its own, including a sex-specific pattern in which female MANF-deficient Purkinje cells showed elevated ER stress markers and degeneration after alcohol exposure that male cells did not.

In both studies, the platform choice was deliberate. Visium found the DEG signal at whole-tissue scale, Xenium (where used) resolved it at single-cell scale, and iPathwayGuide™ turned the resulting gene lists into pathway-level biology.

iSCanGuide™ was built to simplify spatial data analysis. Through guided workflows built on industry best practices, researchers can easily identify differentially expressed genes prior to pathway analysis, without relying on manual methods or separate tools.

iSCanGuide™ accepts raw spatial data directly, with full support for Visium, Xenium, and CosMx platforms, covering the sequencing-based and imaging-based approaches described above in the same environment used for single-cell RNA-seq. From that raw data, iSCanGuide™ delivers spatial domain detection, spatially variable genes, tissue architecture mapping, and cell-type identification, then feeds those results directly into iPathwayGuide™’s Impact Analysis methodology for mechanistic pathway interpretation.

The guided workflows of iSCanGuide™ replace the manual data analysis performed by Kolling² and Wen¹ to identify differentially expressed genes from Visium and Xenium output data. Further, iSCanGuide™ outputs transfer seamlessly into iPathwayGuide™ as part of one integrated workflow for deep pathway and functional analysis.

Imaging-based and sequencing-based platforms answer different questions, and increasingly, the strongest studies use both. What shouldn’t change based on that choice is the path from raw spatial data to pathway-level biology. To see how iSCanGuide handles data from either platform, or to think through which approach fits a specific tissue and question, visit our spatial learning center, browse use cases, or reach out to our team.

Imaging-based methods label a targeted gene panel with probes and image transcripts directly inside cells, capturing spatial location at the moment of detection. Sequencing-based methods capture RNA from spatially barcoded spots on a slide, then sequence it off-tissue and computationally map reads back to their spot of origin. The former trades transcriptome breadth for direct, high-resolution localization; the latter trades resolution for broader, often whole-transcriptome coverage.

Among the platforms discussed here, imaging-based methods such as MERFISH and Xenium reach subcellular to single-cell resolution, while standard sequencing-based platforms like Visium operate at the spot level, with each spot typically spanning one to several cells.³

The choice comes down to the question. If the project needs unbiased, whole-transcriptome differential expression across a tissue section, Visium’s sequencing-based approach is the stronger fit. If the project needs single-cell or subcellular resolution for a defined gene panel, such as mapping cell-cell contacts or rare cell populations, Xenium’s imaging-based approach is the stronger fit. Several recent studies, including Wen et al.¹, use both in the same project: Visium for discovery, Xenium for single-cell confirmation.

Yes. Imaging-based platforms like Xenium, CosMx, and MERFISH are built to resolve individual cells or subcellular structures directly. Standard sequencing-based platforms like Visium capture RNA from multi-cell spots rather than individual cells, though higher-resolution variants such as Visium HD narrow that gap considerably.

Reviewed by Beverly V. Robinson, PhD (Genetics & Molecular Biology, Emory University), Bioinformatics Scientist at Advaita