JavaScript is disabled in your browser. Please enable JavaScript to view this website.

Deep Visual Proteomics (DVP): Connecting Spatial Imaging to Protein Biomarkers

Deep Visual Proteomics (DVP) has immense potential to transform how researchers study protein biomarkers by linking what they see under the microscope to what happens at the molecular level. Traditional proteomics can overlook important signals hidden within complex tissues, while DVP pinpoints specific cells, structures and disease-associated regions to reveal more precise protein signatures. This article explains how DVP works, where it can be applied and what challenges lie ahead before its full integration into biomarker discovery and translational research.

Key Takeaways

  • Deep Visual Proteomics links tissue imaging with proteomics, enabling protein analysis of image-defined cells and tissue regions rather than bulk samples
  • Spatially resolved protein measurements help uncover biomarkers that may be diluted or missed in conventional proteomic workflows
  • AI-assisted image analysis, laser microdissection and mass spectrometry work together to connect cellular phenotypes with protein expression patterns
  • DVP is particularly valuable in oncology, neuroscience, immunology and drug discovery, where cell location and tissue architecture influence biological outcomes
  • While DVP can identify highly promising biomarker candidates, independent validation using complementary methods remains essential before clinical application

What Is Deep Visual Proteomics (DVP)?

Deep visual proteomics (DVP) is an emerging technology that can bridge a longstanding gap between tissue imaging and molecular analysis. Researchers select cells or tissue regions based on microscopy images and link those visual features to protein measurements. DVP commonly profiles pooled cells that share an image-defined phenotype; analysis of individual cells is possible in workflows optimized for single-cell measurements.2,9

By preserving both spatial location and cellular morphology, DVP provides a powerful way to understand how specific cell states, disease-associated features and rare cellular populations are reflected in the proteome. This allows scientists to move beyond simply identifying proteins towards connecting molecular changes to the precise biological landscape in which they occur.2

Spatial information is integral to understanding how cells function within their microenvironments. By linking tissue morphology and histopathological features with protein expression, researchers can identify patterns specific to particular cell types, tissue regions or disease-associated structures. These patterns are often difficult to detect when proteins are measured only from bulk tissue samples.1

Unlike conventional proteomics, which often analyzes mixed cell populations and loses information about cellular localization, DVP maintains a direct link between tissue architecture and protein expression. Within the broader spatial proteomics landscape, DVP distinctively combines image-guided cell selection with phenotype-driven protein profiling, making it especially valuable for biomarker discovery, pathology research and studies of cellular heterogeneity.2

The three types of proteomics are compared and contrasted below.

Aspect
Conventional Proteomics
Spatial Proteomics
Deep Visual Proteomics
Core focus
Measuring proteins across bulk samples or cell populations
Mapping protein expression within tissues
Connecting visually defined cellular phenotypes to protein signatures
Spatial context
Typically lost during sample preparation
Preserved across tissue regions or cellular neighborhoods
Preserved and directly integrated with morphological features
Typical resolution
Bulk tissue or pooled cells
Regional, cellular or near-single-cell depending on the platform
Typically pooled image-selected cells with a shared phenotype or defined tissue regions; individual-cell profiling in optimized workflows
Best suited to
Global protein profiling and pathway analysis
Understanding tissue organization and molecular landscapes
Discovering biomarkers linked to specific cell states, pathological features and rare cell populations

See how Danaher Life Sciences can help

Talk to an expert

Single-Cell and Pooled-Cell Analysis in DVP

Single-cell proteomics is the measurement of proteins at the individual-cell level rather than in pooled cell populations. This approach is most useful when characterizing rare cell populations that contribute disproportionately to disease progression or tissue function within heterogeneous samples.2

In DVP, however, image-guided selection of individual cells does not necessarily mean that each cell is measured separately. Depending on study goals and sample availability, researchers often pool selected cells with similar phenotypes to improve proteome depth, increase measurement robustness and enhance reproducibility. Workflows optimized for single-cell measurements can instead investigate protein differences among individually selected cells in images. This flexibility allows investigators to balance spatial resolution with analytical sensitivity.2,9

When Deep Visual Proteomics Is Most Useful

Deep Visual Proteomics (DVP) is particularly valuable when conventional bulk protein analysis may mask biologically important differences between cell populations. Particularly in heterogeneous tissues, signals from rare, diseased or functionally distinct cells can become diluted when proteins are measured across an entire sample. DVP helps overcome this challenge by linking protein measurements to specifically selected cells or tissue features identified through imaging.3

The approach is especially powerful when tissue morphology, cellular location or interactions within the local microenvironment are central to the research question. For example, researchers investigating tumor margins, immune cell infiltration, fibrotic regions or other visually distinct tissue structures can directly associate these phenotypes with their underlying protein expression profiles while maintaining spatial context.3

From this perspective, DVP is most suitable when researchers need to preserve information about the location and appearance of cells within the tissue.

How Does Deep Visual Proteomics Work?

Deep visual proteomics combines advanced tissue imaging, artificial intelligence (AI), laser microdissection, mass spectrometry and computational analysis into a workflow that links visual tissue features with protein-level information.

A central technical constraint is the very small amount of protein available from microdissected cells. This makes DVP a low-input proteomics workflow: minimizing sample loss during collection and preparation is as important as sensitive protein measurement. Limited starting material can restrict the number of proteins detected, particularly those present at low abundance. Pooling cells with a shared image-defined phenotype can increase protein input and improve proteome coverage, but the resulting measurement represents the pooled group rather than individual cells.

Tissue Collection and Sample Preparation

A DVP workflow begins with the collection and preparation of tissue samples. Depending on the study design, researchers may work with fresh-frozen tissues, formalin-fixed paraffin-embedded (FFPE) specimens, surgical resections, biopsies or other biological samples compatible with imaging and downstream proteomic analysis.4

Sample preservation and processing decisions at this stage are paramount to the rest of the workflow. More specifically, factors such as fixation method, storage conditions, tissue integrity, and section thickness influence protein preservation, tissue morphology and the ability to generate reliable proteomic measurements. On the other hand, poor preservation or excessive tissue degradation may reduce protein recovery and complicate interpretation of results.4

Before imaging begins, sectioning and staining strategies must balance several competing requirements. The tissue must retain sufficient morphological detail for accurate image analysis while remaining compatible with laser microdissection and mass spectrometry.5

Well-characterized clinical specimens, such as archival and FFPE tissues, are among the compatible sample types. However, not every sample type or preparation protocol is equally suitable and study objectives, tissue characteristics and analytical requirements should guide method selection.2

High-Resolution Tissue Imaging

Once tissue sections have been prepared, they are imaged with high-resolution microscopy platforms that capture detailed structural information. The purpose of this stage is to create a digital representation of the tissue that can present morphological characteristics, spatial organization and cellular architecture.6

AI-Assisted Image Analysis and Cell Identification

The digital images are subsequently analyzed using computational image-analysis tools, which often incorporate AI/ML-based methods, helping identify cells, classify tissue features and detect phenotypes. The primary goal of this stage is to convert visual information into quantifiable biological features that can guide targeted proteomic sampling.7

Laser Microdissection of Regions or Single Cells

Following image analysis, laser microdissection is used to isolate the cells or tissue regions selected during imaging. This enables researchers to capture image-defined material for downstream proteomic analysis while maintaining a direct connection to the original tissue.3

Because this targeted approach separates regions of interest from surrounding tissue, it reduces the influence of unrelated cell types. The resulting protein measurements are more closely linked to the selected phenotype or tissue structure, which increases the likelihood of detecting biologically meaningful differences and potential biomarkers.3,10

Mass Spectrometry-Based Protein Identification

The isolated material is then processed for mass spectrometry-based proteomics analysis. During this stage, proteins are extracted, prepared for analysis and measured to identify and quantify the proteins found in the selected cells or tissue regions. This stage generates comprehensive protein profiles with correlations to the image-defined phenotype identified earlier in the workflow.8

Bioinformatics and Spatial Data Integration

After protein measurements are generated, bioinformatics workflows integrate proteomic results with the imaging data acquired earlier in the process. This step reconnects molecular information to the spatial and morphological data from which the samples originated. The resulting integrated dataset helps researchers to explore how protein expression varies across cells, tissue compartments or disease-associated structures.6

Data Analysis and Biological Interpretation

The final analytical stage focuses on converting integrated datasets into biological insights. Researchers perform protein identification and quantification analyses, apply normalization procedures and conduct quality-control assessments to ensure that observed differences reflect biology rather than technical variation.6

One of the insights generated is protein abundance, which can be compared across cell populations, tissue regions, experimental conditions or disease states. Differential protein abundance analysis is often combined with pathway and functional enrichment analyses to identify biological processes associated with specific phenotypes.6

Researchers ultimately examine how observed protein signatures relate to:

The table below summarizes the key inputs, outputs and quality considerations for the imaging, image analysis, microdissection, mass spectrometry and data integration stages.

Workflow stage
Key inputs
Key outputs
Quality considerations
High-resolution tissue imaging

Prepared and stained tissue sections

Imaging protocols optimized for tissue type and staining method

High-resolution digital images

Maps of tissue structures and cellular features

Sufficient resolution, consistent illumination and focus

Artifacts, uneven staining or poor tissue quality can reduce cell identification accuracy

AI-assisted image analysis

High-resolution tissue images

Segmentation and classification algorithms with user-defined phenotypic criteria

Identified cells, cell populations and segmented tissue regions

Phenotypic classifications and image-derived measurements

Accurate segmentation of adjacent cells and tissue structures

Algorithm performance validated against biological expertise, with accuracy affected by staining quality, tissue complexity and image resolution

Laser microdissection

Image-selected cells or tissue regions

Microdissection coordinates generated from image analysis

Isolated cell populations, tissue compartments or individual cells

Material suitable for proteomic processing

Precise alignment between imaging and microdissection datasets

Collection efficiency and sample handling affect protein recovery, and careful selection limits contamination from neighboring cells

Mass spectrometry

Microdissected cells or tissue regions

Protein extraction and preparation reagents, and mass spectrometry instrumentation

Protein identifications and quantitative abundance measurements

Proteomic datasets for downstream analysis

Preparation efficiency directly affects protein recovery, and low starting material may limit proteome coverage

Instrument sensitivity, reproducibility and calibration influence data quality

Bioinformatics and spatial data integration

Protein abundance data

Imaging-derived annotations, spatial coordinates and metadata

Integrated spatial-proteomic datasets

Protein maps linked to tissue locations, in structures suitable for statistical and biological analysis

Accurate mapping between proteomic and imaging datasets

Consistent metadata and validated computational workflows support reproducibility and reduce integration errors

How Deep Visual Proteomics Improves Biomarker Discovery

From Cell-Specific Protein Signatures to Biomarker Candidates

One of the key advantages of Deep Visual Proteomics (DVP) is its ability to measure protein expression within cell types of interest rather than across entire tissue samples. This helps reveal molecular signals that may be diluted in bulk tissue analyses, particularly when the cells of interest represent only a small fraction of the sample.7

By focusing on specific cellular phenotypes, researchers can identify protein signatures associated with particular disease states or rare cells. These cell-specific protein patterns often provide a more precise starting point for biomarker discovery because they are directly linked to the biological features under investigation.11

Because DVP preserves information about tissue morphology, cellular location and cell identity, protein candidates can be assessed not only by their abundance but also by their association with specific cell types or pathological features.2

Researchers can compare spatially resolved protein signatures across samples to determine whether observed patterns remain consistent. Biomarkers that consistently localize to the same cell populations or tissue regions may provide stronger evidence of disease relevance than markers identified solely by bulk analysis.1,4

Where Deep Visual Proteomics Is Applied

Oncology and tumor microenvironment research

Cancer tissues often contain a complex mixture of tumor cells, immune cells, stromal cells and blood vessels, requiring researchers to isolate and profile these distinct cellular populations separately. That is why DVP is invaluable for oncology research, especially for characterizing and comparing different cell types in the tumor microenvironment.

For example, researchers can compare:

These comparisons can help identify proteins associated with tumor progression, metastatic potential, immune evasion and treatment response, supporting the discovery of more precise oncology biomarkers.

Drug discovery and translational research

DVP can help researchers understand how candidate therapies affect specific cell populations within tissues rather than measuring only overall tissue responses. Pharmaceutical researchers may use DVP to:8,16,17

By linking treatment-induced protein changes to visually defined cellular phenotypes, DVP can support target validation and mechanism-of-action studies during drug discovery and development.

Neurodegenerative Disease Research

As with cancer, neurological disorders often involve highly localized pathological changes that are difficult to study with bulk tissue analysis, making DVP highly desirable for investigating specific neuronal populations and their relationships with surrounding tissue.

Immunology and Immune Cell Profiling

The function of immune cells often depends on their location, activation state, and interactions with neighboring cells. DVP offers an emerging approach to investigating these relationships through image-guided protein profiling. Using DVP, researchers can:

Ultimately, DVP can improve our understanding of immune responses and identify biomarkers associated with inflammation, autoimmunity or immunotherapy responses.

Challenges and Current Limitations

Due to the complexity of the advanced methods used in the workflow, DVP faces technical and analytical challenges that may limit its scalable adoption.

Errors may occur as early as sample preparation. Differences in tissue preservation, section quality, fixation methods and staining protocols can introduce variability into imaging and proteomic datasets. This variability may disrupt image analysis performance and complicate comparisons across samples. These risks underscore the need for standardized sample-preparation workflows to ensure reliable results.6,23

DVP relies heavily on computational image analysis to identify cells and classify tissue features. However, errors in cell segmentation or phenotype classification can lead to selecting incorrect cells or regions, ultimately affecting downstream proteomic results. Complex tissue architecture, stacked cells and subtle morphological differences can further complicate accurate identification.23

Among the advanced technologies, DVP combines high-resolution imaging, image analysis, laser microdissection and mass spectrometry. While highly informative, this workflow can be more time- and resource-intensive than conventional bulk proteomics approaches. This holds especially when processing large numbers of samples or extensive clinical cohorts, which require substantial instrumentation, computational resources and technical expertise.10

Finally, like other proteomics and high-content imaging approaches, DVP can be affected by batch-to-batch variation. Differences in sample preparation, staining procedures, microdissection sessions, instrument performance or data acquisition conditions may introduce technical variation that can be mistaken for the underlying biology. These risks can be mitigated not only by standardizing sample preparation and experimental design, but also by implementing quality control measures.6,23

Overall, although deep visual proteomics has emerged as a promising method for biomarker identification, it alone does not sufficiently establish clinical utility. Therefore, candidate biomarkers typically require validation using independent methods, such as immunohistochemistry (IHC), immunofluorescence, targeted proteomics or analyses in larger patient cohorts. As a result, DVP should be viewed as a powerful biomarker prioritization tool that helps identify candidates warranting further investigation rather than a standalone validation platform.

See how Danaher Life Sciences can help

Talk to an expert

FAQ's

How does laser microdissection support deep visual proteomics?

Laser microdissection (LMD) allows researchers to isolate specific cells or tissue regions identified through imaging. Instead of analyzing an entire tissue sample, DVP can target precisely defined cell populations, disease-associated structures or rare cells for protein analysis. This reduces interference from neighboring cells and helps ensure that measured proteins are directly linked to the selected phenotype, improving the specificity of biomarker discovery.

Why is spatial context important in proteomics?

Spatial information reveals where proteins are expressed within a tissue and which cells they are associated with. This helps researchers distinguish meaningful biological signals from background noise, identify region-specific protein patterns and better understand interactions between cells and their microenvironment. As a result, protein findings can be interpreted in a more biologically relevant way.

What technologies are required for deep visual proteomics?

DVP combines several technologies, including high-resolution microscopy, AI-assisted image analysis, laser microdissection, mass spectrometry and bioinformatics software. Together, these tools help researchers identify cells of interest, isolate them from tissue sections, measure their protein content and connect molecular findings back to tissue morphology.

What are the advantages of deep visual proteomics over conventional proteomics?

Unlike conventional proteomics, which averages signals across many cells, DVP analyzes proteins from image-defined cell populations or tissue regions. This allows researchers to detect cell-specific protein signatures, study rare cell types, preserve tissue architecture and identify biomarker candidates that might be missed in bulk analyses.

What are the biggest challenges in Deep Visual Proteomics workflows?

Key challenges include accurate cell segmentation, variability in tissue preparation and staining, limited throughput and technical batch effects. Because the workflow combines multiple advanced technologies, it also requires specialized expertise and careful quality control. In addition, candidate biomarkers identified through DVP still require validation using independent methods.

How does artificial intelligence (AI) contribute to Deep Visual Proteomics?

AI helps analyze complex tissue images by identifying cells, classifying phenotypes and detecting patterns that may be difficult to assess manually. These tools improve the speed, consistency and scalability of image analysis, helping researchers select biologically relevant cells or regions for proteomic profiling while reducing subjectivity in the workflow.

References

  1. Wu M, Tao H, Xu T, Zheng X, Wen C, Wang G, et al. Spatial proteomics: unveiling the multidimensional landscape of protein localization in human diseases. Proteome Sci. 2024;22(1):7.
  2. Mund A, Coscia F, Kriston A, Hollandi R, Kovács F, Brunner AD, et al. Deep Visual Proteomics defines single-cell identity and heterogeneity. Nat Biotechnol. 2022;40(8):1231-1240.
  3. Zheng X, Mund A, Mann M. Deciphering functional tumor-immune crosstalk through highly multiplexed imaging and deep visual proteomics. Mol Cell. 2025;85(5):1008-1023.e7.
  4. Kabatnik S, Post F, Drici L, Bartels AS, Strauss MT, Zheng X, et al. Spatial characterization and stratification of colorectal adenomas by deep visual proteomics. iScience. 2024;27(9).
  5. Kabatnik S, Zheng X, Pappas G, Steigerwald S, Padula MP, Mann M. Deep visual proteomics reveals DNA replication stress as a hallmark of signet ring cell carcinoma. NPJ Precis Oncol. 2025;9(1):37.
  6. Nimo J, Fritzsche S, Valdes DS, Trinh M, Pentimalli T, Klingeberg M, et al. OpenDVP: an experimental and computational framework for Community-Empowered deep visual proteomics. Nat Methods. 2026;1-13.
  7. Min J, Schweizer L, Zonderland G, Selvanesan BC, Thomsen JH, Oldenburg L, et al. AI-powered Deep Visual Proteomics reveals critical molecular transitions in pancreatic cancer precursors. Cancer Discov. 2026;16(7):1323-1340.
  8. Rosenberger FA, Mädler SC, Thorhauge KH, Steigerwald S, Fromme M, Lebedev M, et al. Deep Visual Proteomics maps proteotoxicity in a genetic liver disease. Nature. 2025;642(8067):484-491.
  9. Post F, Hausmann A, Kabatnik S, Steigerwald S, Brand A, Clement DL, et al. Deep visual proteomics reveals an in vivo-like phenotype of orthotopically transplanted human colon organoids. Cell Syst. 2025;16(9).
  10. Guo W, Hu Y, Qian J, Zhu L, Cheng J, Liao J, et al. Laser capture microdissection for biomedical research: towards high-throughput, multi-omics, and single-cell resolution. J Genet Genomics. 2023;50(9):641-651.
  11. Wetzel S, Mann SP, Rosenberger FA. From Single Cells to Diagnosis: Proteomics Technologies in the Multi-Omics Landscape of Rare and Mitochondrial Diseases. J Inherit Metab Dis. 2026;49(5):e70236.
  12. Chakrabarti S, Makhmut A, Mohammadi A, Luo W, Wang L, Lewin GR, et al. Deep visual proteomics uncovers nociceptor diversity and pain targets. Nat Commun. 2026.
  13. Wang Y, Wu Y, Zhang F, Periasamy P, You H, Goh D, et al. Integrating spatial proteogenomics in cancer research. Adv Sci (Weinh). 2026;13(29):e20744.
  14. Zhou XMM, D’Amiano AJ, Lu C, Madan V, Khoshniyati S, Kollings J, et al. Spatial proteomics for investigating solid tumor resistance mechanisms. Cancer Metastasis Rev. 2025;44(4):76.
  15. Zhang Y, Zhu Y, Liu Z, Li H, Zhong B, Cui X, et al. Deep spatial proteomics reveals a suppressive immune niche linked to immune evasion in renal cell carcinoma. Cancer Drug Resist. 2026;9:19.
  16. Meissner F, Geddes-McAlister J, Mann M, Bantscheff M. The emerging role of mass spectrometry-based proteomics in drug discovery. Nat Rev Drug Discov. 2022;21(9):637-654.
  17. Egebjerg K, Chelebian E, Baeksgaard L, Bjerregaard JK, Garbyal RS, Pfeiffer P, et al. Nationwide HER2 Testing, Prospective Trastuzumab-Based Therapy, and Deep Visual Proteomics in Esophageal Squamous Cell Carcinoma. medRxiv. 2026:2026.10.14.26365244.
  18. Ryu T, Kim K, Asiimwe N, Na CH. Proteomic Insight Into Alzheimer's Disease Pathogenesis Pathways. Proteomics. 2025;25(21-22):214-229.
  19. Wang F, Sun C, Wu TW, Fu Y, Fan Y, Zhao S, et al. iPEX enables micrometre-resolution deep spatial proteomics via tissue expansion. Nature. 2026;649(8096):505-514.
  20. Hornburg D, Foser S, Müller T. Deciphering proteome alterations in neurodegeneration with mass spectrometry. Neuroscience. 2025.
  21. Nordmann TM, Anderton H, Hasegawa A, Schweizer L, Zhang P, Stadler PC, et al. Single-cell spatial proteomics identifies the JAK/STAT pathway as an actionable therapeutic target in lethal cutaneous drug reactions. medRxiv. 2023:2023.11.11.23295492.
  22. Zheng X, Pedersen LM, Bzorek M, Mund A, Gjerdrum LMR, Mann M. Deep visual proteomics unveils precision medicine insights in composite small lymphocytic and classical Hodgkin lymphoma. bioRxiv. 2024:2024.06.12.598635.
  23. Siemes D, Novak A, Thiebes S, Widera L, Tautges S, Voss H, et al. Bioimage analysis in deep visual proteomics: Advancing transparency, reproducibility, and FAIR principles. J Microsc. 2026.