Current Advances in Spatial Pathology of Nasal Polyps: Challenges, Opportunities for Artificial Intelligence, and Future Directions Authors: Zi‐Xuan Hua, Xin Luo, Ning Kang, Ze‐Zhi Guo, Man‐Hou Chao, Hua‐Mei Yan, Wan‐Xin Zhang, Tong Wu, Zi‐Qi Zhang, Xue‐Kun Huang, Ya‐Na Zhang, Zhao‐Hui Shi, Qin‐Tai Yang Published: 2026 ABSTRACT ABSTRACTChronic rhinosinusitis with nasal polyps (CRSwNP) is characterized by inflammatory heterogeneity, epithelial hyperplasia, and tissue remodeling, with spatial pathology critical for deciphering pathological mechanisms and guiding clinical practice. Conventional histopathological techniques rely on subjective visual evaluation, limited by inadequate spatial resolution of molecular markers and imprecise quantitative analysis. Recent advances show that spatial transcriptomics reveals nasal polyps (NP) inflammatory heterogeneity; deep learning enables automated inflammatory subset identification and precise endotype prediction; artificial intelligence (AI)‐integrated frameworks transform assessment of tissue remodeling, angiogenesis, and fibro‐inflammatory niches. AI‐enabled spatial pathology bridges morphological and molecular signatures, refines NP endotyping, and paves the way for targeted biologic therapies and precision care. This review summarizes progress in conventional and emerging spatial pathology approaches for NP, focusing on AI‐enabled tools and their clinical translation potential.Keywords: artificial intelligence, chronic rhinosinusitis, digital pathology, nasal polyps, spatial pathology Introduction Chronic rhinosinusitis (CRS) is the most common chronic upper airway disease in otorhinolaryngology, characterized by persistent nasal and sinus mucosal inflammation, with an estimated prevalence of approximately 10% [1]. As a highly heterogeneous disease, it is classified into CRS with nasal polyps (CRSwNP) and without (CRSsNP). Over half of CRSwNP patients experience postoperative recurrence [2], severely impacting quality of life and imposing substantial socioeconomic burdens [3, 4, 5]. Nasal polyps (NP) are characterized by inflammatory cell infiltration, tissue remodeling, and subepithelial edema, with etiologies such as allergies and infections shaping infiltrating cell types and fibrosis extent [6, 7]. Studies investigating CRS pathogenesis have employed various staining techniques, but NP microenvironment heterogeneity involves not only inflammatory cell or cytokine differences but also spatial distribution of angiogenesis, glandular secretion, and fibrosis [8, 9, 10]. Research confirms significant spatial heterogeneity of immune cells in NP subtypes [11, 12, 13], which is critical for elucidating pathogenesis. Transcriptomics provides a holistic perspective for exploring CRS molecular mechanisms [14]. However, limits in spatial resolution, lack of true three‐dimensional (3D) reconstruction, and difficulties integrating artificial intelligence (AI) and multi‐omics into clinical practice still hinder precision diagnosis and treatment. Deep learning (DL), a subset of AI, excels at high‐dimensional biomedical data analysis and has shown preliminary potential in NP pathological classification and inflammatory cell quantification [15, 16]. This paper systematically discusses the current state of research on NP pathology in the epithelial and lamina propria layers, along with the opportunities and challenges of machine learning‐driven multidimensional NP pathology research. A schematic overview of the spatial pathology of NP and the integration of emerging technologies is presented in Figure 1. Schematic overview of the spatial pathology in nasal polyps. Pathophysiology of Nasal Mucosa The nasal mucosa serves as a barrier interface between the external environment and the host immune system. It is mainly composed of the epithelial layer and the lamina propria, whose close interaction maintains mucosal homeostasis and barrier function [6]. In NP, the major pathological alterations include inflammatory cell infiltration, tissue remodeling, and edema, particularly within the subepithelial lamina propria [7]. Rather than occurring as isolated histological findings, these changes are organized across epithelial and stromal compartments, reflecting the local immune microenvironment and tissue response. Various etiologies, such as allergies, infections, and local irritants, can lead to differences in the types of infiltrating cells and the extent of fibrosis [7], further supporting the need to understand NP pathology in a spatially resolved manner. Pathological Research Progress in NP Epithelial Changes Epithelial remodeling is a major pathological feature of NP and is commonly manifested by goblet cell hyperplasia, squamous metaplasia, and abnormalities of pseudostratified ciliated columnar epithelium [17, 18]. Because these epithelial cells are central to mucociliary clearance (MCC), structural and functional ciliary injury can contribute to impaired pathogen clearance and persistent mucosal inflammation. Microbial pathogens such as Pseudomonas aeruginosa and Haemophilus influenzae, together with recurrent infection and inflammatory damage, can disrupt ciliary integrity and coordination, resulting in ciliary loss, fusion, and disordered arrangement [19]. These epithelial abnormalities are among the early pathological events in NP and may not be uniformly distributed across the nasal cavity. However, there is a lack of research comparing the morphological characteristics of ciliated epithelium from different anatomical regions of the nasal cavity. In this context, emerging imaging approaches such as optical coherence tomography (OCT) may provide complementary tools for dynamic and region‐specific assessment of ciliary function [20, 21]. Pathological Changes in the Lamina Propria The lamina propria is a major site of edema, inflammatory cell infiltration, and tissue remodeling in NP, and these changes are shaped by cytokine networks and the local immune microenvironment [22]. Mast cells, eosinophils, T and B lymphocytes, and dendritic cells collectively contribute to the inflammatory cascade, while their interactions with stromal components further promote tissue damage and remodeling [23, 24]. In parallel, fibrosis‐related changes, including collagen fiber degradation, basement membrane thickening, endothelial proliferation, and angiogenesis, are closely linked to vascular abnormalities and local edema [25, 26]. Cytokines and signaling pathways, including interleukin (IL)‐6, vascular endothelial growth factor, transforming growth factor beta 1, and IL‐37, further regulate these inflammatory and remodeling processes [25, 26, 27, 28, 29]. Taken together, the lamina propria should be understood not only as a site of inflammatory infiltration, but also as a spatially organized microenvironment in which immune responses, vascular changes, and stromal remodeling coexist. Future integration of single‐cell sequencing and spatial transcriptomics (ST) may help resolve these local interactions more precisely and provide further insights into inflammation and fibrosis in NP. Analysis of 3D Spatial Features A major limitation of conventional pathological assessment in NP is that a single tissue section captures only part of a spatially heterogeneous lesion. The distribution of inflammatory cells in CRS varies under different inflammatory states, and chronic inflammatory infiltration and fibrosis may extend into deeper tissues [30, 31]. Moreover, pathological assessment based only on inflammatory cell infiltration reflects tissue characteristics from a single dimension, whereas vascular infiltration, epithelial features, and the distribution of glands and blood vessels are also closely related to CRS prognosis [32, 33]. Because these histological components are unevenly distributed across different regions of NP, a single pathological slide may not adequately represent the overall inflammatory state of the polyp [33]. Therefore, comprehensive extraction of spatial features from pathological slides is needed to better characterize NP pathology and to provide a morphological basis for subsequent advances in spatial pathology and three‐dimensional analysis. Mechanism Research Driven by New Technologies Since its inception in the 19th century, microscopic imaging technology has become a critical tool in pathological research. With its ability to analyze at single‐cell resolution, this technology has laid an important foundation for understanding the relationship between tissue structure and disease mechanisms [34]. Although histomorphological analysis remains an essential basis for the diagnosis, prognosis evaluation, and treatment decisions of CRS, directly correlating tissue structural features with functional abnormalities still presents significant challenges. The mechanism research of CRS is gradually shifting toward the molecular pathology level. With the revolutionary advancements in genomics, next‐generation sequencing (NGS), as a core technology in high‐throughput detection systems, has demonstrated significant clinical translational value in the field of head and neck malignant tumors [35]. However, traditional bulk sequencing technologies are unable to resolve cellular heterogeneity, limiting their ability to analyze the heterogeneity of complex clinical diseases. The rise of single‐cell omics and spatial omics technologies has provided new approaches to address these issues. Single‐Cell Sequencing and ST The mature application of single‐cell RNA sequencing (scRNA‐seq) technology provides a direct tool for accurately analyzing the functional roles of various cell types in physiological and pathological processes, helping to reveal cellular heterogeneity in CRS‐related tissues [36, 37, 38, 39]. However, this technology lacks research on pathological features in different regions, requiring more comprehensive analysis relying on ST. ST can integrate gene expression data with spatial location information, reveal disease‐related cellular heterogeneity and molecular mechanisms, and clarify the tissue‐specific mechanisms of CRS treatment resistance and disease recurrence. The comprehensive analysis of spatial transcriptomic data helps deepen the understanding of the overall biological mechanisms of CRS. The summary of CRS pathology studies is shown in Table S1. Inflammation Spatial Pathology Research Driven by New Technologies Whole‐slide imaging (WSI) has enabled the digitization of pathology slides and supported the construction of large‐scale datasets for deep learning model development [40]. Deep learning algorithms have shown broad applicability in the analysis of digitized hematoxylin and eosin (HE) images, including histological classification of cancer [41], prostate cancer grading [42], and prediction of hormone receptor status in breast cancer [43]. However, although HE‐based image analysis captures rich morphological information, its diagnostic and predictive performance remains lower than that of immunohistochemistry (IHC) in some settings. This gap highlights the need to connect histological morphology with molecular information. In this context, joint analysis of ST and HE images provides an important opportunity, because it links local gene expression patterns to tissue architecture. Previous studies have shown that such integration can improve the prediction of local gene expression, enhance detection resolution [44], and strengthen the biological interpretability of model outputs [45]. Despite the rapid development of digital pathology, histological examination remains the gold standard for tissue diagnosis and biological interpretation. A major methodological challenge in ST analysis is that current approaches do not fully exploit the rich morphological information contained in pathology images, and often use histology mainly for spatial localization or simple comparison of cellular similarity [46]. This limits the ability of current models to align tissue architecture with molecular readouts in a biologically meaningful way. In particular, the integration of WSI and ST requires bridging differences in scale, resolution, and information content between image‐based morphology and spot‐based transcriptomic measurements. At the same time, digital pathology and deep learning‐based microscopy image analysis have developed rapidly, and recent pathology foundation models have improved the extraction of cellular and tissue‐level image features [47]. Because ST datasets are generated together with high‐resolution tissue pathology images, they provide a natural basis for cross‐modal integration. On this basis, Ge et al. [48] proposed the GIST framework as a general strategy for combining histology with ST analysis. By incorporating histological information into ST interpretation, this framework improves the accuracy and reliability of ST analysis and offers a more integrated view of tissue spatial structure and disease mechanisms. Although this framework has been validated in lung cancer, breast cancer, and colorectal cancer, its application to non‐cancerous tissues may still face limitations, indicating that further adaptation may be needed for inflammatory diseases such as CRS. In the clinical diagnosis of CRS, HE histological examination remains the core pathological method, whereas the routine clinical application of molecular biomarkers is still limited. In this setting, ST datasets offer a unique advantage by establishing spatial links between gene expression and tissue morphology, thereby providing valuable resources for training next‐generation deep learning models. Such multimodal frameworks may help extend molecular‐level insight from a limited number of profiled samples to broader and more accessible pathological assessment. For CRS pathology research, this creates an important opportunity to move beyond descriptive morphology toward mechanism‐related and prediction‐oriented models. At the same time, successful clinical translation will depend on how effectively future models address the alignment between histological structure and molecular signals in non‐neoplastic inflammatory tissues. AI Empowered 3D Pathology Research on NP In the analysis of nasal polyp pathological mechanisms, 2D spatial analysis techniques, although of significant value, are inadequate when faced with the typical 3D structural features of nasal polyp tissue. There is an urgent need to establish a 3D spatial analysis system to overcome the current research bottlenecks. At present, spatial biology research is primarily based on 2D technology platforms, mainly involving the preparation of 5 μm to obtain tissue morphological or spatial molecular information. Although this technological system plays a key role in analyzing tissue microenvironments, it has significant dimensional limitations: for example, a 5 μm tissue section from a 1 cm nasal polyp biopsy only accounts for 0.05% of its entire volume. This results in 2D analysis providing only discrete spatial pathological information, unable to construct a complete spatial pathological map. Critically, the 3D spatial conformation of NP not only determines their physical properties but also regulates gene expression profiles, cell signaling networks, and multicellular interaction patterns through stereoeffects. In 2D section analysis, the complex spatial structural relationships within the 3D microenvironment are often deconstructed into flat projections, leading to the loss of key biological information. This dimensional reduction may cause cognitive biases on three levels: first, it leads to a systematic underestimation of tissue heterogeneity; second, it fails to accurately reconstruct the 3D pathways of intercellular communication; and third, it over‐simplifies the molecular‐space‐functional coupling mechanisms in the pathological microenvironment, ultimately affecting the systematization of the disease's pathogenesis. Therefore, establishing a multi‐omics integration analysis system based on 3D spatial omics technologies, combined with deep learning‐assisted spatial panoramic reconstruction techniques, will be an important research direction to break through the dimensional limitations in nasal polyp pathology research. 3D pathology technology, as an emerging technique, has demonstrated enormous potential in the diagnosis and treatment of tumors such as prostate cancer [49, 50]. The 3D reconstruction technology based on continuous sections aligns 2D spatial data into the same coordinate space and computationally reconstructs the 3D tissue structure. Recent studies have confirmed the feasibility of this method [51, 52]. To precisely capture morphological features along the z‐axis, new computational models based on continuous section HE images need to be developed for high‐throughput scanning and intelligent analysis. Wu et al. [16] constructed an AI‐assisted quantitative pathology analysis system based on WSI, enabling quantifiable diagnostics of CRS cell phenotypes and allowing rapid, efficient reading of whole‐slide 2D full data, which provides the possibility for 3D pathology research on NP. In previous research, our team utilized the CRS Intelligent Pathological Subtype Diagnostic System‐3D algorithm to rebuild 3D NP, exploring the relationship between inflammatory cells and regional location. By analyzing the full‐slide results of consecutive nasal polyp sections from the root to the base, we identified the optimal biopsy site [53]. 3D technology‐based pathology research on NP has provided a more accurate and comprehensive observational perspective on their pathological features. It has not only significantly enhanced the ability to detect spatial distribution patterns of inflammatory cells and tissue structures that are easily overlooked in thin tissue sections but also facilitated a deeper analysis of tissue characteristics and structural complexity. This has provided a new research perspective for uncovering the pathogenesis of NP and exploring therapeutic targets. Conclusion Research methods, including transcriptomics and spatial biology, have enabled us to better elucidate the mediators and immune pathways involved in the disease progression, thereby deepening our understanding of their role in CRS. From 2D to 3D, spatial inflammatory pathology analysis provides detailed and comprehensive biological information. Notably, a standard 5μm section represents only 0.05% of a 1 cm biopsy volume, highlighting the limited sampling coverage of conventional 2D pathology and further supporting the value of 3D assessment. By integrating 3D spatial omics with AI algorithms, establishing a panoramic analysis framework for nasal polyp 3D pathology, we aim to break through the cognitive boundaries of traditional 2D analysis, offering theoretical breakthroughs in uncovering the 3D pathological mechanisms of NP and discovering precision treatment targets. This not only opens up a new dimension for in‐depth CRS research but also provides a more feasible solution for clinical precision medicine. However, the challenge of how to safely establish a spatial inflammatory pathology data network for CRS and efficiently translate research findings into clinical applications remains to be solved. Inflammation spatial pathology, through the integration of multidimensional technologies, offers a novel perspective for analyzing the heterogeneity of NP. Today, further technological innovation and clinical collaboration are needed to accelerate the translation from basic research to precision medicine, which is expected to improve patient outcomes. In the future, with the continuous development of multidimensional NP pathology research, we will be able to extract more comprehensive and in‐depth CRS pathological information, uncovering many new diagnostic and therapeutic targets, thus providing new ideas for personalized diagnosis and treatment of CRS. Author Contributions Z.X.H. and X.L. drafted the manuscript and conducted the literature review. N.K., Z.Z.G., M.H.C., H.M.Y., W.X.Z., T.W., and Z.Q.Z. contributed to literature collection, data interpretation, and manuscript revision. X.K.H., Y.N.Z., and Z.H.S. gave constructive comments. Q.T.Y. conceived and supervised the study and critically revised the manuscript. Ethics Statement This manuscript is a literature review and involved no human subjects. Therefore, ethical approval and informed consent were not required. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File Acknowledgments The authors have nothing to report. This work was supported by the National Key R&D Programmes Subproject (2022YFC2504100); General Program of the National Natural Science Foundation of China (82501384, 82271148, and 82571289); The Key Area Research and Development Programme of Guangdong Province (2023B1111040004); Sun Yat‐sen University Clinical Research 5010 Program (2019006); “Five by Five” Project of the Third Af liated Hospital of Sun Yat‐sen University (2023WW601); Young Talent Support Project of Guangzhou Association for Science and Technology (No. QT‐2025‐035); China Postdoctoral Science Foundation under Grant Number (No. GZC20242098); Key Laboratory of Airway In ammatory Disease Research and Innovative Technology Translation, Guangzhou (2023A03J0209); Key development Program of the Institute of Arti cial Intelligence, Sun Yat‐sen University (2025RGZN006); Special Project on Climate Change of China Meteorological Administration (QBZJ202502); Guangdong Provincial Young Scientific and Technological Talent Cultivation Project of Guangdong Association for Science and Technology (2026QNRC373). Hua Z.‐X., Luo X., Kang N., et al., “Current Advances in Spatial Pathology of Nasal Polyps: Challenges, Opportunities for Artificial Intelligence, and Future Directions,” World Journal of Otorhinolaryngology ‐ Head and Neck Surgery 12 (2026): 546‐552. 10.1002/wjo2.70145. Zi‐Xuan Hua and Xin Luo contributed equally to this study. Data Availability Statement Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study. References Associated Data Supplementary Materials Supporting File Data Availability Statement Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.