- In situ hybridization (ISH) and imaging-based transcriptomics are important approaches in spatial molecular biology that allow researchers to detect RNA molecules directly within intact cells and tissue sections. Unlike conventional RNA sequencing, which generally requires RNA extraction and loses the original spatial arrangement of cells, these techniques preserve information about where specific transcripts are located. They help researchers investigate gene expression in relation to tissue architecture, cellular identity, developmental processes, disease conditions, and interactions between neighbouring cells. In bioinformatics, the resulting images and molecular measurements are processed computationally to identify transcripts, quantify gene expression, segment cells, and reconstruct spatial patterns of gene activity.
- In situ hybridization is based on the principle of complementary base pairing between a labelled nucleic acid probe and its target RNA sequence. A probe is designed to recognize a specific transcript, and when it hybridizes with the complementary RNA molecule in a fixed tissue or cell, the target can be detected through a fluorescent, chromogenic, or other signal. The tissue is generally fixed and permeabilized to preserve its structure and allow probes to enter cells. Hybridization conditions are optimized to promote specific binding, followed by washing steps to remove unbound or nonspecifically bound probes. The resulting signal reveals the presence and location of the target transcript.
- Traditional chromogenic in situ hybridization uses enzyme-based reactions to produce a coloured signal that can be observed under a conventional light microscope. Fluorescence in situ hybridization (FISH) uses fluorescently labelled probes to detect RNA molecules under a fluorescence microscope. RNA-FISH is particularly useful for examining the cellular and subcellular distribution of specific transcripts, including their localization within the nucleus, cytoplasm, and other cellular regions. Multiplexed FISH extends this approach by detecting multiple RNA targets in the same tissue section through different fluorescent labels, sequential hybridization, or combinations of imaging and signal-decoding strategies.
- Several advanced techniques have expanded the capabilities of in situ hybridization. RNAscope uses paired oligonucleotide probes and signal amplification to improve the sensitivity and specificity of RNA detection. It is useful for detecting low-abundance transcripts and validating gene-expression patterns identified through RNA sequencing. Single-molecule FISH (smFISH) uses multiple fluorescent probes that bind to different regions of a target transcript, allowing individual RNA molecules to be visualized as distinct fluorescent spots. This enables researchers to estimate transcript abundance at the single-cell level and investigate differences in gene expression between individual cells. These targeted methods are especially valuable for validating candidate genes, studying rare transcripts, and examining spatial expression patterns.
- Imaging-based transcriptomics extends these principles to the measurement of many RNA species within intact tissues. Instead of examining only one or a few transcripts, advanced imaging-based platforms use multiplexed detection strategies to identify hundreds or thousands of RNA targets. These methods generally involve repeated cycles of probe hybridization, fluorescence imaging, signal decoding, and, where necessary, probe removal or signal resetting. Each transcript is identified through a specific combination or sequence of fluorescent signals. High-resolution microscopy captures the signals, and computational algorithms decode them to determine the identity and location of individual RNA molecules.
- Multiplexed Error-Robust Fluorescence In Situ Hybridization (MERFISH) is an important imaging-based transcriptomics technology. It uses combinatorial fluorescent barcodes to identify many RNA species in individual cells. Each RNA target is assigned a unique barcode, which is detected through multiple rounds of hybridization and imaging. Error-correcting coding strategies help identify transcripts even when some signals are missing or incorrectly detected. MERFISH can provide spatially resolved information about thousands of genes, depending on the implementation and assay design, and is particularly useful for studying cellular organization, tissue architecture, and gene-expression heterogeneity.
- Sequential Fluorescence In Situ Hybridization (seqFISH) and seqFISH+ use repeated rounds of fluorescent probe hybridization and imaging to identify large numbers of RNA transcripts. Different combinations of fluorescent signals across imaging cycles produce unique molecular codes. These methods allow researchers to measure gene expression in individual cells while retaining spatial information. They are useful for investigating complex tissues, including brain regions, developing organs, and tumour microenvironments. Their performance depends on probe design, imaging quality, signal detection, and the accuracy of computational decoding.
- Commercial imaging-based platforms have also made spatial transcriptomics more accessible. 10x Genomics Xenium uses in situ hybridization and fluorescence imaging to detect selected RNA targets within tissue sections at cellular and subcellular resolution. NanoString CosMx Spatial Molecular Imager uses multiplexed molecular imaging to measure RNA and, in compatible workflows, protein targets within individual cells. These platforms provide spatial coordinates for detected molecules and support the analysis of cellular organization and molecular relationships. Their gene coverage is generally determined by predefined probe panels, although panel design and available assay configurations may vary.
- The computational analysis of in situ hybridization and imaging-based transcriptomics involves several interconnected stages. First, microscopy images are acquired and assessed for focus, illumination, background fluorescence, tissue quality, and signal consistency. Image preprocessing may include background subtraction, illumination correction, denoising, image registration, and alignment of images from different hybridization cycles. Image registration is particularly important in multiplexed methods because images captured during separate cycles must be accurately aligned to ensure that signals are assigned to the correct spatial locations.
- The next major step is transcript detection and decoding. Computational algorithms identify fluorescent spots or other molecular signals within images and determine which RNA targets they represent. In MERFISH and seqFISH, decoding algorithms interpret combinations of signals across imaging rounds to identify individual transcripts. Error correction can help manage missing or incorrectly detected signals. Detected molecules are then assigned spatial coordinates, producing a molecular dataset in which each transcript is associated with its position in the tissue.
- Cell segmentation is another essential component of imaging-based transcriptomics. Segmentation identifies individual cell boundaries and distinguishes nuclei, cytoplasm, and surrounding tissue regions. Nuclear staining, membrane markers, cytoplasmic signals, and computational image-analysis methods can be used to define cellular boundaries. Detected transcripts are subsequently assigned to cells according to their spatial positions. This produces a cell-by-gene expression matrix that can be used for downstream analysis. Incorrect segmentation can lead to transcript misassignment, particularly in densely packed tissues, overlapping cells, or regions with poorly defined boundaries.
- A range of computational tools supports the analysis of imaging-based transcriptomics. Platform-specific software includes Xenium Ranger for processing and analysis of Xenium data, while commercial imaging platforms provide their own processing and visualization environments. General image-analysis tools such as Fiji/ImageJ, CellProfiler, and QuPath can support image processing, cell segmentation, and quantitative analysis. StarDist and Cellpose are widely used for deep-learning-based cell and nuclear segmentation. Python libraries such as OpenCV, scikit-image, NumPy, and SciPy support image processing and computational analysis. Seurat, Scanpy, Squidpy, Giotto, and SpatialExperiment can be used for downstream expression analysis, visualization, spatial neighbourhood analysis, and integration with other spatial or single-cell datasets.
- After transcript detection and cell assignment, researchers perform quality control and expression analysis. Important quality indicators include the number of detected transcripts per cell, the number of genes detected, background signal levels, the proportion of unassigned molecules, cell segmentation quality, and the consistency of signal detection across imaging cycles. Low-quality cells, damaged tissue regions, and unreliable molecular detections may need to be excluded or investigated separately. Gene-expression profiles can then be used for cell-type identification, clustering, differential expression analysis, and characterization of cellular states.
- Imaging-based transcriptomics can also be integrated with single-cell RNA sequencing. Single-cell RNA sequencing provides broad transcriptome information but usually loses the original spatial location of each cell. Imaging-based methods retain spatial organization and can help map selected genes to specific cells and tissue regions. Reference mapping and cell-type annotation can combine single-cell reference profiles with spatial measurements. Tools such as Seurat, Scanpy, Tangram, and other spatial integration methods can support this process. However, differences in gene coverage, sensitivity, tissue processing, and measurement technology must be considered when interpreting integrated datasets.
- An important application of imaging-based transcriptomics is the investigation of cellular neighbourhoods and cell–cell communication. By identifying the spatial positions of cells and their gene-expression profiles, researchers can examine which cell types occur close to one another and investigate potential molecular interactions. Ligand–receptor analysis tools such as CellChat, CellPhoneDB, NicheNet, and LIANA can help identify candidate communication pathways when appropriate expression data are available. Spatial neighbourhood analysis can also reveal the organization of immune cells, stromal cells, epithelial cells, and other populations within complex tissues. These computational predictions require biological interpretation and, where possible, experimental validation.
- In situ hybridization and imaging-based transcriptomics have broad applications in cancer research, neuroscience, developmental biology, immunology, infectious disease research, and regenerative medicine. In cancer, they can identify tumour-associated gene-expression patterns, immune-cell distributions, and molecular differences between tumour regions. In neuroscience, they help map neuronal subtypes and the spatial organization of gene expression across brain structures. In developmental biology, they reveal how gene expression changes across embryonic tissues and during organ formation. In animal genomics and breeding research, these methods can be used to study gene expression in muscle, mammary glands, reproductive organs, immune tissues, and other biologically important tissues. They can help investigate cellular mechanisms underlying growth, milk production, fertility, disease resistance, and adaptation to environmental conditions.
- Despite their advantages, in situ hybridization and imaging-based transcriptomics have several limitations. Traditional ISH and targeted FISH methods usually examine a limited number of genes, while advanced multiplexed methods require more complex probe designs, imaging systems, and computational processing. Fluorescence background, tissue autofluorescence, probe accessibility, RNA degradation, incomplete hybridization, and imaging errors can affect detection accuracy. Cell segmentation is particularly challenging in densely packed tissues, where cellular boundaries may overlap or be difficult to distinguish. Imaging-based methods may also have lower transcript detection efficiency than expected, and the absence of a detected signal does not necessarily mean that a gene is not expressed.
- The reliability of results can be improved through careful probe design, appropriate positive and negative controls, optimized tissue preparation, standardized imaging procedures, and independent validation. RNA-FISH and RNAscope can be used to validate candidate transcripts identified through RNA sequencing, while immunofluorescence and immunohistochemistry can provide complementary information about protein localization. Combining imaging-based transcriptomics with spatial proteomics, single-cell RNA sequencing, histology, and functional experiments can provide a more complete understanding of tissue organization and molecular regulation. As imaging technologies, probe chemistry, and computational methods continue to improve, in situ hybridization and imaging-based transcriptomics are becoming increasingly important approaches for studying gene expression at cellular and subcellular resolution.