The framework, described in a study published in PLOS Computational Biology by Dee Velazquez, Caleb Hallinan, Roujin An, Kalen Clifton, and Jean Fan, addresses a fundamental blind spot in imaging-based spatially resolved transcriptomics, a family of technologies that measure gene expression with molecular resolution while preserving the physical layout of tissue. These platforms can record the precise coordinates of thousands of individual mRNA molecules across millions of cells, producing datasets of extraordinary richness. Yet the standard analysis pipeline throws most of that richness away. By collapsing the molecular map into a gene-by-cell count matrix, conventional methods reduce each cell to a bag of genes, discarding the spatial texture that the imaging platforms worked so hard to capture.

STARIT’s central insight is deceptively simple: instead of converting transcripts into counts, convert them into pixels. The method takes all of the mRNA molecules detected within the boundary of a segmented cell and rasterizes them into a standardized, image-based tensor representation. Each gene, or set of genes, becomes a channel of the tensor, and the spatial distribution of transcripts within the cell becomes the pixel values of that channel. The result is a stack of images—one per gene or gene set—that encodes not just what a cell is expressing but where, inside the cell, each transcript resides. Because the representation is a tensor of fixed structure, it plugs directly into the deep learning computer vision models that have transformed fields from medical imaging to autonomous driving.

This design choice matters because it lets researchers borrow the full arsenal of modern computer vision. Convolutional neural networks, which excel at detecting spatial patterns in images, can now be trained to recognize spatial patterns in gene expression. A network trained on STARIT tensors can learn, for example, that transcripts of a particular gene tend to hug the nuclear envelope in one cell state and disperse toward the cell membrane in another. Such patterns are invisible to count-based methods by construction, since both configurations would produce identical tallies. By preserving geometry, STARIT opens a dimension of cellular characterization that has been technically accessible but computationally neglected.

The authors put their framework through a rigorous evaluation using both simulated and real imaging-based spatial transcriptomics data. In simulations, where the ground truth is known by design, STARIT could distinguish transcriptionally distinct cell types with high fidelity, matching the performance expected of conventional approaches. The real test, however, was whether it could go further. Across real datasets, the method not only separated known cell types but also distinguished cell states defined purely by subcellular transcript localization—states that conventional gene count analysis failed to capture. In other words, two cells with the same gene expression profile could be told apart if their transcripts were arranged differently inside them, a capability that no bag-of-genes representation can offer.

The implications reach into some of the most active areas of biology and medicine. Cell state, rather than cell type, often determines how a tissue responds to disease. Tumor cells shift into invasive, drug-tolerant, or proliferative states; immune cells flip between activated and exhausted configurations; neurons remodel their local molecular architecture in response to activity. If the spatial arrangement of transcripts within cells tracks these transitions, then a method that reads that arrangement could detect disease-relevant states earlier and more reliably than count-based classifiers. Subcellular localization is also intimately tied to biology itself: mRNA molecules are actively transported to specific cellular neighborhoods, where local translation supports functions like cell migration, polarity, and synaptic signaling. A framework that quantifies localization at scale turns a qualitative observation into a measurable, learnable feature.

Part of STARIT’s significance lies in its standardization. Imaging-based spatial transcriptomics has suffered from a proliferation of bespoke analysis approaches, each lab rasterizing, encoding, and modeling its data differently, which makes results hard to compare and reproduce. By defining a common tensor format for subcellular molecular information, STARIT gives the field a shared substrate on which models can be built, benchmarked, and transferred. A model trained on one tissue or platform could, in principle, be fine-tuned on another, much as pre-trained vision models are adapted across image domains today. That kind of interoperability could accelerate a shift from one-off analyses to cumulative, community-wide model development.

The framework also highlights a broader conceptual shift underway in computational biology: the migration of biological data into representations suited to deep learning. Just as protein structures became representable as geometric graphs amenable to neural networks, and histology slides became pixel grids for diagnostic AI, the interior of the cell is now becoming an image. Each of these transitions unlocked capabilities that the previous data formats could not support. STARIT suggests that the transcriptome’s spatial texture is the next frontier, and that the tools of computer vision—already battle-tested on natural images—may be repurposed with relatively little modification to mine it.

Challenges remain, as they do for any young method. Rasterizing transcripts into images requires choices about resolution, channel organization, and normalization, and the authors’ demonstration that the approach works on simulated and real data is a first step rather than a final verdict. Scaling to tissues with millions of cells and tens of thousands of genes will demand careful engineering, and interpreting what a vision network has learned about subcellular biology will require new explainability tools. Still, the direction is clear. By refusing to average away the molecular geography of the cell, STARIT gives biologists a way to see cell states that were always there but never before systematically measurable.

For a field that has spent two decades perfecting the art of counting molecules, the message of this study is that counting was only ever half the picture. The other half is written in the positions of those molecules, in patterns of nuclear clustering and cytoplasmic dispersal that encode how a cell is organized and, quite possibly, what it is becoming. With STARIT, that half of the picture is finally rendered in a form machines can learn from—and the cells hiding in plain sight within our tissues may soon give up secrets that gene counts alone could never reveal.

Subject of Research: A computational framework that represents subcellular transcript localization in imaging-based spatial transcriptomics as image tensors for deep learning-based cell state characterization

Article Title: Spatial Transcriptomics As Rasterized Image Tensors (STARIT) characterizes cell states with subcellular molecular heterogeneity

Article References: Velazquez, D., Hallinan, C., An, R., Clifton, K., & Fan, J. (2026). Spatial Transcriptomics As Rasterized Image Tensors (STARIT) characterizes cell states with subcellular molecular heterogeneity. PLOS Computational Biology, 22(10), e1014845. https://doi.org/10.1371/journal.pcbi.1014845

Image Credits: AI Generated

DOI: 10.1371/journal.pcbi.1014845

Keywords: spatial transcriptomics, STARIT, subcellular localization, deep learning, computer vision, cell states, single-cell analysis, computational biology, gene expression, image tensors, PLOS Computational Biology, mRNA

The research team turned to spatial transcriptomics, a technology that measures gene expression across intact tissue sections while preserving the physical coordinates of every measurement. Unlike single-cell RNA sequencing, which dissociates tissue and loses positional information, spatial transcriptomics allows researchers to ask not only which genes are active, but where. The investigators applied this approach to tumor samples containing TLS and used a computational framework called EcoTyper to infer cell states — transcriptionally coherent subpopulations within broader cell lineages — directly from the spatial data. In total, they identified 71 distinct cell states spanning all of the major immune lineages present in the tumor microenvironment, from B cells and T cells to dendritic cells and neutrophils.

This cell state-level resolution proved to be the study’s decisive methodological choice. When the researchers examined how each inferred state was distributed along a gradient running from the geometric core of a TLS to its surrounding periphery, they found that cells belonging to the same lineage could behave in dramatically different ways. Some lineages contained states that were enriched within the TLS and others that were depleted there, meaning that a coarse analysis lumping all cells of a lineage together would have averaged away the very patterns the researchers were seeking. Core-periphery organization, the authors conclude, is in an important sense invisible at the cell type level and only emerges at the cell state level.

Among the most notable discoveries were two TLS-enriched CD4 T cell states with clearly distinct marker gene programs. Although both belonged to the helper T cell lineage and both accumulated within tertiary lymphoid structures, they carried different transcriptional signatures, suggesting they perform different functions within the immune hub. One of these states, designated CD4.T_S02, was more abundant toward the geometric core of the TLS, while its counterpart occupied more peripheral positions. This kind of fine-grained spatial segregation within a single lineage hints at a division of labor inside TLS that mirrors, in miniature, the zoned organization of canonical lymphoid organs such as lymph nodes and the spleen.

To complement the abundance-gradient analysis, the team constructed weighted chemokine networks centered on three key signaling molecules: CXCL13, CXCL12, and the paired chemokines CCL19 and CCL21. Chemokines are the trafficking signals of the immune system, guiding cells to their destinations through gradients and receptor-ligand matching. CXCL13 and CCL19/21 are classically associated with lymphoid organization — they recruit B cells and helper T cells into structured aggregates and help maintain the boundary between distinct lymphoid zones — while CXCL12 has been implicated in both lymphoid organization and immunosuppressive niches. By mapping how these chemokine signals relate to the inferred cell states across space, the researchers could ask whether the transcriptional picture of TLS organization was corroborated by an independent, signaling-centered lens.

It was. The chemokine network analysis revealed two recurrent communities of cell states. The first was a lymphocyte activation community confined to the TLS themselves, consistent with the idea that the core of these structures is where immune cells are being primed and activated. The second was an effector-focused community that extended outward into the surrounding tumor tissue, suggesting that TLS do not merely house immune activity but actively export it — dispatching armed effector cells into the tumor proper, where they can encounter and attack malignant cells. This core-to-periphery functional gradient provides a mechanistic rationale for the clinical association between TLS presence and better outcomes: a tumor with well-organized TLS is not just decorated with immune cells, it is plugged into a production line for antitumor immunity.

One of the most intriguing recurring characters in this spatial drama was a neutrophil cell state of tentative identity that appeared as a hub across all three chemokine axes the researchers examined. Neutrophils are usually cast as short-lived, blunt instruments of innate immunity, and in many tumors they are associated with immunosuppression. The repeated emergence of a neutrophil state at the center of TLS-associated chemokine networks raises the possibility that a specialized neutrophil population may play an architectural or regulatory role within these structures — perhaps helping to organize the recruitment of lymphocytes, or modulating the signals that keep the TLS intact. The authors are careful to label this state’s identity as tentative, and defining it precisely will be a task for future work, but its centrality across independent analyses makes it one of the study’s most provocative leads.

Taken together, the two analytical strategies — the TLS-centric abundance gradient and the weighted chemokine networks — converge on a coherent model. Tertiary lymphoid structures, in this view, are spatially zoned organs: an active core where lymphocyte activation and germinal center-like processes dominate, surrounded by a more diffuse effector periphery where armed immune cells transition into the tumor tissue. The zones are not arbitrary; they are organized through a shared, key set of chemokine signaling pathways, with CXCL13, CXCL12, and CCL19/21 acting as the molecular mortar that holds the architecture in place. This zoned model reframes TLS from static histological features into dynamic, functionally compartmentalized machines for generating and deploying antitumor immunity.

The implications for cancer medicine are considerable. TLS are already the subject of intense interest as biomarkers for immunotherapy response, and numerous clinical efforts aim to induce or mature them in tumors that lack them. The new findings suggest that such efforts should be evaluated not simply by whether TLS appear, but by whether they acquire the correct internal organization — the right cell states in the right places, wired together by the right chemokine signals. A disorganized aggregate of lymphocytes may not function like a properly zoned TLS, and therapies that reshape chemokine signaling could, in principle, convert poorly structured infiltrates into fully functional immune hubs. The cell state atlas generated in this study offers a reference against which such interventions could be measured.

The study, conducted under ethics approvals from the University of Tokyo and Keio University School of Medicine and supported by JSPS KAKENHI grants, also demonstrates the maturing of spatial transcriptomics as a discovery platform. By combining computational state inference with explicit spatial modeling and network analysis, the researchers extracted organizational principles that neither approach alone would have revealed. As the technology spreads and atlases of this kind accumulate across cancer types, the field moves closer to a genuinely spatial understanding of the tumor immune microenvironment — one in which the arrangement of cells, not merely their presence, becomes a target of measurement, prediction, and ultimately therapy. For now, the message of this work is clear: to understand how tumors’ hidden immune organs work, one must look not just at who is there, but at who is where, and at the chemical conversations that put them there.

Subject of Research: Spatial organization and chemokine network architecture of tertiary lymphoid structures in tumors

Article Title: Spatial transcriptomics reveals cell state-level organization and chemokine network architecture within tertiary lymphoid structures

Article References: Yan, A., Iwasawa, T., Tanaka, N., Kakimi, K., Lysenko, A., & Tsunoda, T. (2026). Spatial transcriptomics reveals cell state-level organization and chemokine network architecture within tertiary lymphoid structures. Cancer Immunology, Immunotherapy. https://doi.org/10.1007/s00262-026-04529-2

Image Credits: AI Generated

DOI: 10.1007/s00262-026-04529-2

Keywords: tertiary lymphoid structures, spatial transcriptomics, chemokine signaling networks, cell states, tumor microenvironment, cancer immunology, CXCL13, CXCL12, CCL19, CD4 T cells, neutrophils, immunotherapy biomarkers