Resolution, Sensitivity, and Throughput: How Spatial Platforms Differ

Every vendor leads with a different headline number, and those numbers rarely mean the same thing from one spatial transcriptomics platform to the next. Spatial transcriptomics resolution, detection sensitivity, gene plex, and throughput are the four specifications that actually distinguish one system from another. Understanding how each is measured turns a marketing spec sheet into a genuine basis for comparison.

Defining resolution in spatial transcriptomics

Spatial transcriptomics resolution describes the smallest area of tissue that a platform can assign to a distinct expression measurement, and it varies by roughly two orders of magnitude across current technologies. Resolution makes up one of the four key spatial transcriptomics metrics, as aforementioned (Figure 1). Sequencing-based platforms originally operated at the scale of a tissue spot rather than a single cell, with early Visium arrays from 10xGenomics capturing RNA within 55-micrometer (µm) spots that typically span several cells at once.

Figure 1: A four-panel overview of the metrics that distinguish spatial transcriptomics platforms: resolution, sensitivity, gene plex, and throughput. Credit: AI-generated image created using Google Gemini (2026).

Newer sequencing-based designs have considerably improved resolution. Stereo-seq, currently provided by STOmics, captures RNA at roughly 0.5 µm, while Visium HD bins data at 2 µm. Both are precise enough to approach single-cell or even subcellular assignment, according to a large subcellular platform benchmark.

On the other hand, imaging-based platforms such as Xenium, also by 10xGenomics, and CosMx by Bruker derive resolution from the physics of single-molecule fluorescence detection. This enables them to localize individual transcripts rather than binning a fixed grid. A dedicated comparison of single-molecule spatial platforms covered this approach in depth. This nanoscale-to-microscale span is one of the central trade-offs researchers weigh across the broader spatial biology field when planning a study.

A platform's advertised resolution is not always its effective resolution in practice. A comprehensive review of spatial transcriptomics technologies frames resolution as one of several methodological dimensions that must be read alongside molecular throughput and transcriptome coverage rather than in isolation. Molecular diffusion during tissue permeabilization can also blur the boundary of a nominal spot or bin, so the figure printed on a spec sheet can overstate the resolution a platform actually achieves in a given tissue.

Spatial sensitivity and detection efficiency explained

Detection sensitivity, sometimes called spatial sensitivity or capture efficiency, is distinct from resolution. It describes the proportion of the RNA molecules physically present in a section that are captured, sequenced, or imaged, and ultimately counted. A platform can offer subcellular resolution while still missing most of the transcripts present if it has low capture efficiency.

Sensitivity varies not only by platform but by tissue type and even by permeabilization time within the same platform. A systematic comparison of sequencing-based spatial transcriptomic methods found that molecular diffusion is a variable parameter across both technologies and tissues. None of the sequencing runs tested reached saturation, even though sequencing depth ranged from roughly 300 million reads on one platform to 4 billion reads on another. In practice, sensitivity depends heavily on how deeply a sample is sequenced. The same study reported that gene-specific capture bias could vary by platform, with certain marker genes reliably detected on one system but largely undetected on another, even within the same tissue section.

This has a practical consequence for experimental design. A platform that appears highly sensitive on one reference tissue may perform quite differently on a more complex or densely packed sample. Sensitivity figures drawn from an unrelated tissue type should therefore be treated as a starting point, not a guarantee.

Imaging-based platforms also track a related property called specificity. Together, sensitivity and specificity make up what researchers often call spatial detection efficiency, and specificity itself is assessed using negative-control probes designed not to bind any real transcript. A six-cancer-type technical comparison reported markedly different background rates between two widely used imaging platforms, illustrating that sensitivity and specificity must be evaluated together rather than as one combined score. A separate three-platform imaging benchmark across dozens of tumor and healthy tissue types found that two of the three platforms resolved slightly more distinct cell subclusters than the third on matched tissue. The authors suggested that differences in resolution reflected factors such as the presence of key marker genes in each platform’s panel. Researchers weighing these trade-offs often start with foundational context on sequencing- and imaging-based methods before comparing individual platform specifications.

Gene plex vs whole transcriptome coverage

Gene plex refers to the number of distinct genes a platform can measure simultaneously, and it is where sequencing- and imaging-based platforms diverge most sharply. Sequencing-based methods capture RNA transcriptome-wide by default, without preselecting a gene list, whereas imaging-based methods must design a targeted probe panel in advance.

That targeting comes with a real cost. The same 6-cancer-type comparison found that expanding one imaging platform's panel by roughly 13-fold, from a few hundred genes to 5,000, did not yield a proportional increase in detected transcripts. Instead, results indicated that per-gene sensitivity decreased as panel size increased. This trade-off was attributed to optical crowding, as probe density increases within a fixed imaging area. Larger panels can still add value by resolving finer-grained cell subtypes, but that benefit comes with sparser data for any individual gene.

Choosing a plex level, then, is a genuine trade-off rather than a simple "more is better" decision. A study benchmarking four high-throughput platforms with subcellular resolution found that those with panels exceeding 5,000 genes still varied widely in the consistency with which their transcript counts matched single-cell RNA sequencing data. Panel size alone, in other words, does not guarantee data quality.

Spatial throughput and imaging area trade-offs

Throughput, sometimes discussed as spatial throughput, describes how quickly a platform can process tissue and how large an area it can practically image or capture in a single run. It interacts directly with both resolution and plex. Higher resolution generally means more data points per unit area; this increases computational and imaging burden and can slow a run considerably.

The imaged area also varies with platform design, not just by chemistry alone. Some imaging-based systems scan a continuous region spanning multiple square centimeters in a single pass, while others capture discrete, user-defined fields of view that must be stitched together afterward. This difference affects both hands-on time and how representative a sampled region is of the whole tissue section.

Circulating Immune Proteins in Liver Cancer Research

This webinar explores how circulating immune-related proteins can improve our understanding of hepatocellular and gallbladder cancer risk across diverse populations with viral hepatitis.
View Webinar

Advertisement

Sampling area has downstream statistical consequences that are easy to overlook. For example, the six-cancer-type comparison found that estimating cell-type composition from a single small field of view yielded highly variable results. In one case, the estimated cancer cell content ranged from roughly 5% to 55%, while the true value for the full scanned region was 22%. Larger or more numerous sampled regions reduced this variability substantially, underscoring that throughput and imaged area shape not just convenience but the reliability of a study's quantitative conclusions.

Researchers planning a study should weigh these four dimensions (Table 1) against the biological question at hand rather than defaulting to whichever number looks most impressive on a vendor page.

• Start with the biological unit that matters for the question, since mapping tissue architecture at the neighborhood level requires far less precision than localizing individual transcripts within a single cell.
• Confirm whether the experiment needs unbiased transcriptome-wide coverage or a smaller, well-validated gene panel focused on known markers.
• Ask what tissue area the study actually requires, since a platform optimized for small regions of interest may be a poor match for whole-section profiling.
• Check whether published sensitivity and specificity figures came from fresh-frozen or formalin-fixed, paraffin-embedded tissue, since performance can differ substantially between the two.
• Look for benchmarking data generated on tissue types similar to the planned study, since platform rankings can shift meaningfully by tissue.

Table 1: A synthesized comparison of the four metrics that define spatial platform performance, drawn from multiple peer-reviewed benchmarking studies.

Reading spatial transcriptomics resolution specs with confidence

Resolution, sensitivity, plex, and throughput each answer a different question about a spatial platform, and a single headline figure never captures all four at once. Reading a spec sheet critically means asking which of these dimensions a vendor is actually reporting and checking whether that figure was measured under conditions similar to those of the tissue and experiment being planned.

Independent, peer-reviewed benchmarking studies remain the most reliable way to compare platforms honestly. They test multiple systems under matched conditions, rather than relying on manufacturer-reported specifications alone. As new subcellular and high-plex platforms continue to enter the field, the underlying framework of resolution, sensitivity, plex, and throughput will remain the clearest lens for evaluating what each one actually delivers.

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.