Mass spectrometry imaging is a rapidly developing spatial molecular analysis technology. It combines the molecular detection capability of mass spectrometry with the spatial representation capability of image analysis, allowing the spatial distribution of molecules to be directly obtained from tissue sections, cell samples, or the surfaces of other biological materials. Unlike conventional mass spectrometry, mass spectrometry imaging does not rely on bulk extraction of the entire sample before analysis. Instead, it scans the sample surface point by point or region by region while preserving the spatial structure of the tissue. Each pixel corresponds to a mass spectrum, and each mass-to-charge ratio signal can be reconstructed into an ion image. Therefore, mass spectrometry imaging can answer not only “what molecules are present” but also “where these molecules are located” [1].

From a historical perspective, the idea of mass spectrometry imaging can be traced back to surface mass spectrometry analysis. The emergence of modern biomedical mass spectrometry imaging is generally associated with the application of MALDI mass spectrometry to molecular localization in tissue sections in the 1990s. In 1997, Caprioli and colleagues reported the use of MALDI-TOF MS for the two-dimensional spatial localization of peptides and proteins in biological samples, which laid an important foundation for molecular imaging of tissues [1]. Subsequently, Stoeckli and colleagues published a study in Nature Medicine that further demonstrated the potential of mass spectrometry imaging for analyzing protein expression in mammalian tissues, bringing this technique into the fields of biomedical research and molecular pathology [2]. Since then, technologies such as MALDI, DESI, SIMS, ion mobility mass spectrometry imaging, and high-resolution mass spectrometry imaging have continued to advance. The detectable targets of mass spectrometry imaging have expanded from proteins to lipids, metabolites, small-molecule drugs, glycans, peptides, and many other classes of molecules. The development of DESI mass spectrometry imaging has enabled researchers to observe the spatial distributions of drugs and their metabolites in tissues under near-ambient pressure conditions with relatively limited sample preparation [3].

The basic workflow of mass spectrometry imaging usually includes tissue collection, cryosectioning or sample preparation, matrix application or in situ desorption/ionization, mass spectrometric acquisition, spectral preprocessing, ion image reconstruction, molecular annotation, spatial segmentation, and subsequent statistical modeling. For MALDI mass spectrometry imaging, the sample surface is usually coated uniformly with a matrix. A laser irradiates predefined spatial grids point by point, causing local molecules to undergo desorption and ionization, after which the mass spectrometer records the intensities of ions with different mass-to-charge ratios. After data reconstruction, researchers can obtain spatial distribution maps of a single molecule or a group of molecules in the tissue, and can further integrate them with pathological images, immunostaining, transcriptomic data, proteomic data, or clinical information. High-resolution mass spectrometry imaging can also be combined with database searching and statistical control methods to improve the reliability of metabolite annotation. The FDR-controlled metabolite annotation method proposed by Palmer and colleagues represents an important advance in moving MSI data from “ion image display” toward “interpretable spatial metabolomics analysis” [4].

One important feature of mass spectrometry imaging is label-free detection. It does not depend on antibodies, fluorescent probes, or predefined targets, and can simultaneously detect a large number of molecular signals. This makes it suitable for discovering unknown or unexpected metabolic alterations. Another important feature is in situ spatial analysis. The distributions of metabolites, lipids, and drugs in tissues often show clear regional differences, which can be obscured by bulk extraction-based analysis. Mass spectrometry imaging places molecular information back into the tissue structure, allowing researchers to observe molecular differences among lesion regions, boundary regions, necrotic regions, immune-infiltrated regions, and normal tissues. It also has strong multiplexed molecular detection capability, as a single experiment can generate thousands or even more ion images, providing a data basis for tissue heterogeneity analysis, molecular subtyping, and disease mechanism studies. Bouslimani and colleagues used mass spectrometry data to construct a three-dimensional molecular map of the human skin surface, showing that MSI and related molecular cartography techniques can be applied not only to tissue sections but also to spatial chemical analysis of complex biological surfaces [5].

The significance of mass spectrometry imaging lies in its ability to provide a spatial molecular perspective for disease research. Many diseases do not occur uniformly throughout the entire tissue, but instead manifest as local changes in the microenvironment, metabolic reprogramming, and differences in cellular states. Metabolic heterogeneity in tumors, region-specific metabolic abnormalities in neurodegenerative diseases, drug distribution and accumulation in tissues, and lipid changes in inflammatory microenvironments all require analysis at the spatial scale. Mass spectrometry imaging can detect molecular changes within a spatial framework close to pathological observation, and is therefore helpful for connecting tissue morphology, molecular mechanisms, and clinical phenotypes. In recent years, mass spectrometry imaging has also begun to move toward single-cell spatial metabolomics. The SpaceM method proposed by Rappez and colleagues combines MALDI-MSI with microscopy images and single-cell spatial features, enabling in situ analysis of metabolic states at the single-cell level and demonstrating the expanding potential of MSI in cellular heterogeneity research [6].

At present, mass spectrometry imaging has been widely used in cancer research, neuroscience, drug development, toxicology, pathological diagnosis, plant and food science, environmental exposure research, and spatial metabolomics. In cancer research, it can be used to analyze metabolic differences within tumors and at tumor margins, and to identify molecular signals associated with invasion, metastasis, drug resistance, or prognosis. In neuroscience, it can be used to observe changes in the distributions of lipids and metabolites in brain regions, providing spatial molecular evidence for studies of Alzheimer’s disease, Parkinson’s disease, and other brain disorders. In drug development, it can be used to investigate the distributions of drugs and their metabolites in tissues, and to evaluate targeted delivery, tissue accumulation, and potential toxicity. In translational clinical research, it provides new technical routes for molecular pathology, surgical margin assessment, and disease subtyping.

Mass spectrometry imaging also faces several key challenges. First, MSI data are high-dimensional, noisy, and often contain many missing values, and systematic differences can easily arise across batches, instruments, and sample preparation conditions. Second, there is often a trade-off between the spatial resolution of ion images and the confidence of molecular annotation. Third, MSI data and pathological images, immunohistochemistry, spatial transcriptomics, and clinical data differ in resolution, coordinate systems, and information types, making direct integration difficult. Therefore, the development of mass spectrometry imaging depends not only on instrumentation and experimental techniques, but increasingly on computational methods, image processing, machine learning, statistical modeling, and multimodal data integration.

Our team’s research interests in mass spectrometry imaging mainly lie at the intersection of computational methods and biomedical applications. Given the high noise level, high dimensionality, limited annotation, and strong spatial heterogeneity of MSI data, our team focuses on how machine learning and deep learning methods can be used to improve the quality, representation capability, and interpretability of MSI data. Related research includes MSI data preprocessing, missing value imputation, denoising, ion image representation learning, spatial segmentation, multimodal image fusion, and spatial heterogeneity analysis. iSegMSI improves the spatial segmentation of MSI data through interactive prior information, demonstrating the value of combining human knowledge with unsupervised models in the analysis of complex MSI data [7]. DeepION focuses on ion image representation, using deep learning to extract low-dimensional spatial features for the identification of colocalized ions and isotope ions [8].

In the field of spatial metabolomics, our team is interested in advancing mass spectrometry imaging from “molecular distribution mapping” to “spatial metabolic mechanism analysis.” This requires linking individual ion images, metabolite sets, tissue regions, cell types, and disease phenotypes to establish an analytical framework that can explain disease progression and changes in the tissue microenvironment. Our research places particular emphasis on multiscale metabolic interaction networks and molecular network modeling, aiming to understand metabolic abnormalities in disease from the perspectives of spatial distribution, metabolic coordination, and network reconstruction. GraphMSI integrates metabolic spectral information with spatial adjacency relationships and uses graph deep learning to analyze spatial heterogeneity in MSI data, reflecting a shift in this field from pixel-level analysis toward spatial structure modeling [9]. Related work can be applied to studies of Alzheimer’s disease, hepatocellular carcinoma, nasopharyngeal carcinoma, gastric cancer, diabetic nephropathy, and other diseases, providing computational tools for disease subtyping, mechanism analysis, and potential biomarker discovery.

In multimodal spatial analysis, our team focuses on the joint analysis of mass spectrometry imaging with histopathology, immune imaging, PET, MRI, spatial transcriptomics, and other types of data. A single technology usually captures only one aspect of a biological system, while metabolic alterations, cellular composition, tissue structure, and clinical phenotypes in diseased tissues are closely and complexly related. Through multimodal registration, image fusion, spatial segmentation, and statistical modeling, information from different levels can be integrated within a unified spatial coordinate system, thereby improving the reliability of spatial molecular interpretation. In this process, mass spectrometry imaging provides direct molecular detection capability, while machine learning and image analysis provide methodological support for cross-modal information integration.

Overall, our team’s research goal in mass spectrometry imaging is to develop computational mass spectrometry and spatial multi-omics analysis methods for complex biomedical problems. On the one hand, we aim to establish more robust data processing and intelligent analysis methods to address the noise, missing values, low signal-to-noise ratio, and high-dimensional features of MSI data. On the other hand, by integrating molecular network modeling, machine learning, and multimodal spatial data fusion, we aim to reveal metabolic heterogeneity and spatial molecular mechanisms in diseased tissues. Through these studies, we hope to promote the development of mass spectrometry imaging from a molecular visualization technology into an interpretable spatial omics analysis tool, providing a new data basis and methodological support for disease mechanism research, molecular subtyping, and precision medicine.

References

[1] Caprioli, R. M.; Farmer, T. B.; Gile, J. Molecular Imaging of Biological Samples: Localization of Peptides and Proteins Using MALDI-TOF MS. Analytical Chemistry, 1997, 69(23): 4751–4760.

[2] Stoeckli, M.; Chaurand, P.; Hallahan, D. E.; Caprioli, R. M. Imaging Mass Spectrometry: A New Technology for the Analysis of Protein Expression in Mammalian Tissues. Nature Medicine, 2001, 7(4): 493–496.

[3] Wiseman, J. M.; Ifa, D. R.; Zhu, Y.; Kissinger, C. B.; Manicke, N. E.; Kissinger, P. T.; Cooks, R. G. Desorption Electrospray Ionization Mass Spectrometry: Imaging Drugs and Metabolites in Tissues. Proceedings of the National Academy of Sciences of the United States of America, 2008, 105(47): 18120–18125.

[4] Palmer, A.; Phapale, P.; Chernyavsky, I.; Lavigne, R.; Fay, D.; Tarasov, A.; Kovalev, V.; Fuchser, J.; Nikolenko, S.; Pineau, C.; Becker, M.; Alexandrov, T. FDR-Controlled Metabolite Annotation for High-Resolution Imaging Mass Spectrometry. Nature Methods, 2017, 14(1): 57–60.

[5] Bouslimani, A.; Porto, C.; Rath, C. M.; Wang, M.; Guo, Y.; Gonzalez, A.; Berg-Lyon, D.; Ackermann, G.; Moeller Christensen, G. J.; Nakatsuji, T.; Zhang, L.; Borkowski, A. W.; Meehan, M. J.; Dorrestein, K.; Gallo, R. L.; Bandeira, N.; Knight, R.; Alexandrov, T.; Dorrestein, P. C. Molecular Cartography of the Human Skin Surface in 3D. Proceedings of the National Academy of Sciences of the United States of America, 2015, 112(17): E2120–E2129.

[6] Rappez, L.; Stadler, M.; Triana, S.; Gathungu, R. M.; Ovchinnikova, K.; Phapale, P.; Heikenwalder, M.; Alexandrov, T. SpaceM Reveals Metabolic States of Single Cells. Nature Methods, 2021, 18(7): 799–805.

[7] Guo, L.; Liu, X.; Zhao, C.; Hu, Z.; Xu, X.; Cheng, K. K.; Zhou, P.; Xiao, Y.; Shah, M.; Xu, J.; Dong, J.; Cai, Z. iSegMSI: An Interactive Strategy to Improve Spatial Segmentation of Mass Spectrometry Imaging Data. Analytical Chemistry, 2022, 94(42): 14522–14529.

[8] Guo, L.; Xie, C.; Miao, R.; Xu, J.; Xu, X.; Fang, J.; Wang, X.; Liu, W.; Liao, X.; Wang, J.; Dong, J.; Cai, Z. DeepION: A Deep Learning-Based Low-Dimensional Representation Model of Ion Images for Mass Spectrometry Imaging. Analytical Chemistry, 2024, 96(9): 3829–3836.

[9] Guo, L.; Xie, P.; Shen, X.; Lam, T. K. Y.; Deng, L.; Xie, C.; Xu, X.; Wong, C. K. C.; Xu, J.; Fang, J.; Wang, X.; Xiong, Z.; Luo, S.; Wang, J.; Dong, J.; Cai, Z. Unraveling Spatial Heterogeneity in Mass Spectrometry Imaging Data with GraphMSI. Advanced Science, 2025, 12(8): 2410840.