For the analytical community, spatial analysis by mass spectrometry is the part of the spatial revolution that gets least attention and covers the most chemical space. While the genomics platforms map transcripts and the immunology platforms map proteins with antibodies, mass spectrometry imaging maps metabolites, lipids, drugs, and peptides directly in tissue — without labels, without probes, and without deciding in advance what to look for.
Key Takeaways
|
What Does Spatial Analysis Mean for Analytical Scientists?
It means giving up the homogenate. Conventional analytical workflows extract, separate, and quantify — and in doing so they average away location. Spatial analysis preserves the tissue and asks where each molecule was, which changes both the instrumentation and the question. A concentration difference between two tumors is a result. A concentration gradient across the margin of one tumor is a mechanism.
The distinguishing feature of the mass spectrometry route is that it is label-free and untargeted. As a review of mass spectrometry imaging for spatially resolved multi-omics mapping in npj Imaging puts it, label-free MSI capitalizes on the intrinsic advantages of mass spectrometry, allowing thousands of molecules to be imaged in a single experiment without prior knowledge and without labels or antibodies. That is a genuinely different proposition from a platform where you order a panel and detect what you ordered.
The Argument That Defines This Territory Ask which spatial technique reaches which molecular class and the answer is uneven in a way that matters. Transcripts can be reached by sequencing or by hybridization probes. Proteins can be reached by antibodies or by mass spectrometry. Metabolites and most lipids can be reached by mass spectrometry and by nothing else, because there is no antibody for lactate and no probe that hybridizes to a phospholipid. That is not a competitive claim about platforms. It is a consequence of chemistry, and it means the entire spatial metabolomics and spatial lipidomics field belongs to analytical science by default. If your question involves small molecules in tissue, the antibody and sequencing platforms are not slower alternatives; they are simply not applicable. |
Molecular Class | Spatial Approaches Available | Practical Position |
mRNA transcripts | Sequencing-based capture, in situ hybridization imaging | Well served by dedicated commercial platforms |
Proteins, targeted | Antibody-based multiplexed imaging, or MS with mass-tagged antibodies | Competitive. Choice depends on plex, throughput, and whether targets are known |
Proteins, untargeted | Mass spectrometry, typically after microdissection | MS only. Antibody methods cannot detect what was not selected |
Peptides and glycans | Mass spectrometry imaging | MS territory in practice |
Lipids | Mass spectrometry imaging | MS only for untargeted coverage |
Metabolites | Mass spectrometry imaging | MS only. No antibody or hybridization route exists |
Drugs and metabolites of drugs | Mass spectrometry imaging | MS only, and the basis of spatial pharmacology |
Elements and inorganics | SIMS, laser ablation approaches | MS territory |
Table 1. Spatial approaches by molecular class. The lower half of the table is why mass spectrometry imaging is not an alternative to the antibody and sequencing platforms but a complement to them.
Working in analytical science?
Register for a FREE Separation Science account to subscribe to the Separation Science Newsletter.
Mass Spectrometry Imaging Fundamentals
The principle is simple to state. Rather than extracting analytes into solution, the instrument samples the tissue surface directly at a defined position, ionizes whatever is there, and records a mass spectrum. Repeat that across a grid and every pixel carries a full spectrum. Selecting any mass-to-charge value and plotting its intensity across the grid produces an ion image, a map of where that species was.
The consequence is that MS imaging is inherently untargeted at acquisition. You are not detecting a chosen analyte; you are recording everything that ionized and deciding later what to look at. That is also why the data is large and awkward, a point taken up further down.
Three ionization approaches do most of the work, and they differ in physics rather than in quality. A society-published beginner's guide to mass spectrometry imaging in The Biochemist sets out the distinctions clearly, and a review of MALDI imaging advances in Frontiers in Chemistry covers the dominant approach in more depth.
| MALDI | DESI | SIMS |
Ionization | Pulsed laser rastered across a matrix-coated surface | Charged solvent spray under ambient conditions | High-energy primary ion beam sputtering the surface |
Environment | Typically vacuum | Ambient, atmospheric pressure | Typically vacuum |
Hardness | Soft. Minimal fragmentation | Soft | Hard. Fragments larger molecules |
Analyte range | Widest: metabolites, lipids, peptides, proteins, glycans, drugs | Metabolites, lipids, some proteins | Elements, inorganics, small organics |
Reported resolution | Commonly single-digit to tens of microns | Limited in part by spray footprint and solvent spreading | Reported from roughly 50 nm to 10 microns |
Sample preparation | Matrix application required, by nebulization or sublimation | Minimal. Analysis in near-native state | Minimal coating, but vacuum-compatible |
Best suited to | Broad untargeted coverage across molecular classes | Rapid mapping with minimal intervention | Subcellular and elemental imaging |
Table 2. The three workhorse ionization approaches for mass spectrometry imaging, compared on characteristics reported in the peer-reviewed and society literature rather than on vendor specifications. Confirm current performance figures for any specific instrument with its manufacturer.
Two nuances are worth carrying forward. MALDI is the most widely adopted approach partly because it produces mainly singly charged ions, which yields simpler spectra than electrospray and makes interpretation more tractable. And DESI’s ambient operation is not merely a convenience: analyzing tissue in its native state without extensive preparation avoids a whole class of preparation artifacts, at some cost in achievable resolution. Variants such as nano-DESI, which uses two capillaries to form a continuous liquid bridge at the surface, push that resolution higher.
The Trade-off You Cannot Design Around Spatial resolution and sensitivity oppose each other, and the reason is physical rather than technical. Halving the pixel dimension quarters the sampled area, and therefore roughly quarters the number of molecules available to ionize at that position. Push resolution far enough and low-abundance species fall below detection. Reviews of high-spatial-resolution MS imaging make the same point from experience: increasing resolution tends to increase experimental complexity and can reduce coverage, sensitivity, or both. So the honest framing for method development is not which instrument has the best resolution, but what resolution your question actually requires, then what coverage you can achieve at that resolution. That framing, and how to choose an operating point deliberately, is developed further in the fundamentals section. |
Each approach rewards more detail than a comparison table can carry, because the differences that matter in practice sit in sample handling and method development rather than in headline specifications. The fundamentals section takes them individually, working through matrix chemistry and laser parameters for MALDI, the preparation artifacts that ambient operation avoids along with the nano-DESI and liquid extraction variants developed to improve its resolution, primary ion beams and depth profiling for SIMS, and the resolution against sensitivity trade-off as a cross-cutting method development question. Start with Mass Spectrometry Imaging: Principles, Techniques, and Applications.
Why Is Mass Spectrometry the Only Route to Spatial Metabolomics?
Because affinity reagents do not exist for most small molecules — and cannot straightforwardly be made. Antibodies are raised against epitopes, and a metabolite of a few hundred daltons typically presents no adequate epitope. Hybridization probes read nucleic acid sequence, which a lipid does not have. The absence is structural, not a gap awaiting a product launch.
This gives spatial metabolomics and spatial lipidomics a different character from the rest of spatial biology. There is no platform to select and no panel to design, because the measurement is untargeted by construction. What you choose instead is ionization approach, polarity, mass range, resolution, and matrix, which are method development decisions of exactly the kind analytical scientists make routinely.
Three consequences follow that are worth stating plainly.
- Annotation replaces target selection as the hard problem. An untargeted experiment returns many detected features, and identifying them from mass alone is genuinely difficult, particularly for isomers. Confidence in annotation, and being explicit about its level, becomes central rather than incidental.
- Quantification is harder than in a homogenate. Ion suppression and matrix effects vary across a tissue section, so signal intensity is not straightforwardly proportional to concentration between regions of differing composition. Quantitative work requires deliberate design.
- Lipids reward the approach disproportionately. Lipids ionize well, are abundant, and are strongly compartmentalized, which makes spatial lipidomics one of the most productive applications of the technique.
Because there is no platform decision to make here, the useful guidance concerns method development and interpretation rather than procurement. The metabolomics and lipidomics section follows a workflow end to end from polarity and mass range selection through to feature detection, examines why lipids are the most productive class and why their isomers are harder to resolve than mass accuracy alone suggests, treats annotation and confidence reporting as a subject to settle before generating a dataset rather than after, and addresses what the word quantitative can defensibly mean when ion suppression varies across a section. It begins with Spatial Metabolomics and Lipidomics by Mass Spectrometry Imaging.
MS-Based Spatial Proteomics
Proteins are the one class where mass spectrometry and antibody methods genuinely compete, and the distinction is the familiar one between targeted and untargeted measurement. An antibody-based platform detects the proteins whose antibodies you applied, at high sensitivity and increasingly high plex. A mass spectrometry approach detects what is present, without that constraint, and can resolve post-translational modifications and proteoforms that an antibody panel is blind to.
Three workflow families dominate, and they answer different questions.
- 1. Direct MS imaging of peptides and proteins. MALDI imaging can map peptides and proteins directly from a section, giving spatial context across the whole tissue at the cost of depth of coverage compared with a solution-phase experiment.
- 2. Microdissection followed by liquid chromatography and tandem MS. Excise a defined region, then run a conventional deep proteomics workflow on it. Spatial resolution is set by the dissection rather than the instrument, and coverage is much deeper. This is the workhorse for region-level questions.
- 3. Imaging-guided microdissection at cell-population level. Use imaging to identify cells of interest, isolate precisely those, then analyze by MS. This is the approach that brings untargeted proteomics close to single-cell spatial resolution.
One point of overlap deserves noting because it confuses newcomers. Mass spectrometry can also be used in a targeted, antibody-dependent mode, by staining tissue with antibodies conjugated to unique mass reporters and reading those reporters by MS imaging. Hundreds of probes can be applied simultaneously. That is a genuine MS technique, but it is targeted, so it sits alongside the antibody platforms conceptually rather than with untargeted MS. Readers coming from a proteomics background may find our existing Proteomics: Key Techniques, Emerging Trends, and Applications guide a useful foundation before going further.
Since this is the one class where the choice between mass spectrometry and antibody methods is genuinely open, the proteomics section covers both the individual workflows and the comparison itself. It works through microdissection coupled to liquid chromatography as the practical workhorse, including the sample handling problems that appear at very low input, the sensitivity limits that govern what is detectable from single-cell material, the imaging-guided approaches that bring untargeted proteomics close to cell-type resolution, and a direct comparison against antibody-based methods on plex, sensitivity, post-translational modifications, and cost. The entry point is Spatial Proteomics by Mass Spectrometry: LCM, Single-Cell, and Imaging Approaches.
Sample Preparation and Data Analysis
These are the two stages where MS imaging experiments actually succeed or fail, and both differ from what a solution-phase analytical scientist is used to.
On preparation, the tissue is the sample, and it cannot be cleaned up. There is no extraction step to remove interferences, no chromatographic separation before detection in most imaging workflows, and no opportunity to adjust concentration. Everything that will be measured is already in place, so the preparation decisions are about preserving spatial fidelity and molecular integrity simultaneously. Section thickness, mounting, washing to remove interfering salts without displacing analytes, and for MALDI the choice and application of matrix all determine what is detectable. Matrix application in particular is a craft: crystal size limits achievable resolution, and uneven application produces signal variation that looks like biology.
On data, an imaging run produces a spectrum per pixel, which is a hyperspectral cube rather than a chromatogram. That brings genuine analytical challenges. The literature on spatial pharmacology is explicit that the high dimensionality of MS imaging data, while rich in information, creates data-analytic difficulty, and that machine learning and deep learning approaches are increasingly applied to extract tissue heterogeneity from it. Multimodal integration, aligning MS imaging with histology and with other spatial modalities, is a distinct problem again and depends on coregistration accuracy rather than on either measurement.
Four practical points for anyone scoping a first experiment:
- Decide resolution before anything else, because it constrains preparation, acquisition time, coverage, and data volume together.
- Plan coregistration with histology from the outset. Retrospective alignment to a section that was not imaged for the purpose is considerably harder.
- Budget acquisition time realistically. A high-resolution raster across a whole section is measured in hours, and sometimes considerably longer.
- Establish how you will annotate features before you generate them, since an untargeted dataset without an annotation strategy is a collection of unidentified masses.
Preparation deserves its own section precisely because it is where the avoidable failures live, and because each decision constrains the ones that follow. That section covers the preservation question, which behaves quite differently for small molecules than it does for nucleic acids; then sectioning thickness and mounting practices that preserve spatial fidelity rather than quietly redistributing analytes, matrix selection and application as the craft step where crystal size sets a practical ceiling on resolution, and coregistration to stained sections, which is what makes an ion map interpretable to a pathologist. It starts with Sample Preparation for Spatial Analysis: From Tissue to Data.
The analysis challenge is different in kind from a chromatographic workflow, and the data section treats processing, interpretation, and application separately. It covers the preprocessing choices such as peak picking and normalization that can substantially shape a reported result, the registration accuracy that determines whether combining a chemical layer with a transcriptional one means anything, unsupervised segmentation and classification together with the interpretability question those raise, and spatial pharmacology, which is the application area that adopted this technique earliest and most seriously. Begin with Analyzing Mass Spectrometry Imaging Data: Processing, Statistics, and Multimodal Integration.
Where Does Mass Spectrometry Fit in the Spatial Landscape?
Alongside the other modalities rather than against them, and the division of labor is reasonably clear once you separate the molecular classes. Sequencing-based and hybridization-based platforms own transcripts. Antibody-based imaging owns high-sensitivity targeted protein detection at scale. Mass spectrometry owns everything small, everything untargeted, and anything where you do not know in advance what you are looking for.
If Your Question Is | The Appropriate Modality | Why |
Which genes are expressed where | Sequencing or in situ hybridization imaging | Direct nucleic acid measurement, mature platforms |
Where are these twenty known proteins | Antibody-based multiplexed imaging | High sensitivity for defined targets |
What proteins are present in this region | Microdissection plus LC-MS | Untargeted, and resolves modifications |
Where is this drug and its metabolites | MS imaging | No alternative modality reaches xenobiotics |
How does metabolism differ across this tissue | MS imaging | No antibody or probe route to metabolites |
What lipid species define these structures | MS imaging | Lipids ionize well and are strongly compartmentalized |
I do not know what I am looking for | MS imaging | Untargeted at acquisition, by construction |
Table 3. Matching the question to the modality. The final row is the honest summary of where mass spectrometry has no substitute.
The most productive work increasingly combines them, using MS imaging to discover which molecules and regions matter and then a targeted modality to characterize those regions further, or aligning both onto the same histology. That integration is a coregistration and data problem more than an instrumentation one, which is why it appears in the data analysis section of this guide rather than the fundamentals.
For analytical scientists, the practical conclusion is encouraging. The skills that spatial analysis by mass spectrometry demands—method development, ionization understanding, matrix effects, annotation discipline, and quantitative rigor—are the skills the analytical community already has. What is new is the requirement to preserve and interpret spatial context, and that is learnable. Further coverage across separation and detection methods is collected in our Omics topic hub.
This article was produced under Separation Science's AI Editorial Guidelines.
