空间转录组技术的快速发展为刻画与解析组织微环境的时空动态交互提供了革命性的工具，使得在单细胞或亚细胞分辨率下解析组织微环境的基因表达空间模式成为可能，为揭示组织发育、稳态维持、疾病进程等生理病理进程的时空动态规律提供了前所未有的数据基础。作为多细胞生命体功能实现的核心机制，细胞间通讯（Cell-Cell Communication, CCC）通过配体-受体信号网络协调细胞行为，在胚胎形态发生、器官结构形成以及免疫微环境调控等多种生命过程中发挥了关键作用。现有的主流空间交互分析方法延续了传统单细胞转录组基于细胞类型进行聚类的分析范式，通过固定半径与近邻细胞类型定义了局部相互作用，从而引入空间邻域的概念以利用局部的空间信息。然而受限于细胞身份的离散化定义与局部空间内相互作用的区域化解析策略，目前的主流方法难以捕捉和反映跨区域的信号梯度、多配体受体的空间竞争等全局性空间特征，缺乏对组织尺度空间整体结构的信息的利用，难以描述全局性空间交互模式。

基于CCC在组织稳态调控中兼具空间运输的方向性与动态连续性的双重特征，我们根据理想各向同性扩散的假设，提出了基于高斯定理和泊松方程的保守力场模型，将配受体相互作用及其空间输运关系映射为类静电场的矢量场分布。通过定义场势、场强和力矢投影分量等物理量，实现了从单细胞分辨率到组织结构区域尺度的多层级空间相互作用的量化，能够进行细胞间、分群间以及区域间的多种相互作用强度比较。

为了验证算法和模型的有效性和健壮性，我们在10x Visium、Xenium、Vizgen MERSCOPE及Stereo-seq等多种主流空间转录组技术所生成的人类/小鼠健康与癌变组织数据集上开展了系统性验证。我们成功发现了FGF家族受体在小鼠的小脑与后脑区域的分布差异所驱动的空间信号竞争；人多阶段食管癌变进展中，EFNB1-EPHB4信号强度的阶段依赖性增强，及其背后的基因动态变化如何促进了肿瘤发生。而在高分辨率数据中，我们验证出CXCL趋化因子家族的空间差异表达在乳腺癌中调控了三级淋巴结构形成，在卵巢癌中诱导了肿瘤内部免疫细胞浸润的减少，并揭示了CXCL相关的CD8 T细胞基因表达模式。最后，我们发现了在 dMMR与pMMR两种分型的结直肠癌之间，由IFNG表达差异所介导的空间细胞相互作用模式如何影响并构建了免疫微环境的异质性。

这些验证实验表明，场论模型及分析算法在Visium、Xenium、MERSCOPE等多种平台中均有稳定表现，能够精细地分辨、识别出空间局部与全局空间的细胞相互作用及其竞争关系，并通过场论模型挖掘相应的生物学意义，揭露其背后的基因表达模式与动态调控等深度信息。该算法广泛适用于不同的空间分辨率与转录本覆盖度的多种技术平台与场景，在解析全局空间相互作用规律方面具有独特优势，为癌症微环境演化、器官发育模式形成等涉及多层级空间交互的生物学问题提供了全新的分析工具。通过引入场论模型，本研究实现了空间转录组数据分析范式的转变：将传统基于局部的离散化描述拓展为连续化的全局系统建模，显著提升空间信息利用率，为解析多尺度空间互作网络提供了新的框架与思路。

The rapid advancement of spatial transcriptomics has revolutionized the characterization and analysis of spatiotemporal interactions in tissue microenvironments, enabling the resolution of spatial expression patterns at single-cell or subcellular scale. This breakthrough provides an unprecedented data foundation for deciphering the spatiotemporal dynamics underlying physiological and pathological processes, including tissue development, homeostasis maintenance, and disease progression. As a core mechanism governing multicellular biological functions, cell-cell communication (CCC) coordinates cellular behaviors through ligand-receptor signaling networks, playing pivotal roles in multiple biological process including embryonic morphogenesis, organogenesis, and immune microenvironment regulation. Current mainstream analytical frameworks for spatial interactions extend the cell-type clustering paradigm established in single-cell transcriptomics, where local interactions are estimated through fixed-radius neighborhood definitions and proximity-based cell type associations to integrate spatial context. However, constrained by discrete cell identity definitions and localized interaction resolution strategies, existing approaches exhibit limited capacity to capture global spatial features such as cross-regional signaling gradients and multi-ligand-receptor spatial competitive dynamics. Due to inefficient integration of global spatial information, those methods fail to characterize global interaction patterns.

To address the vectorial signaling and temporal continuity of CCC, we propose a conservative force field model based on Gauss's theorem and Poisson's equation under ideal isotropic diffusion assumptions. This framework maps ligand-receptor interactions and their spatial transport relationships into electrostatic-like vector fields. By defining physical quantities including field potential, field intensity, and force vector projection components, we achieve multi-scale quantification of spatial interactions from single-cell resolution to tissue-regional levels, enabling comparative analysis of interaction strengths across individual cells, cell clusters, and anatomical regions.

To validate the efficacy and robustness of our algorithm, we conducted systematic evaluations on datasets generated by multiple spatial transcriptomic platforms (10x Visium, Xenium, Vizgen MERSCOPE, Stereo-seq) from human/mouse healthy and cancerous tissues. The key findings include: 1) Spatial signal competition driven by differential FGF receptor distribution in mouse cerebellar and hindbrain regions; 2) Stage-dependent  increasing of EFNB1-EPHB4 signaling intensity during human esophageal carcinogenesis, coupled with dynamic gene expression changes promoting tumorigenesis; 3) Spatial heterogeneity of CXCL chemokine family regulating tertiary lymphoid structure formation in breast cancer and reducing intra-tumoral immune infiltration, with identification of CXCL-associated CD8+ T-cell gene expression patterns in ovarian cancer; 4) IFNG-mediated spatial CCC mechanisms shaping immune microenvironment heterogeneity between dMMR and pMMR colorectal cancers.

These validations demonstrate that our field-theoretic model consistently resolves local and global spatial interactions across platforms, deciphers biological significance through field dynamics, and reveals underlying gene regulatory networks. This algorithm demonstrates broad applicability across diverse technological platforms with varying spatial resolutions and transcriptomic coverages, exhibiting unique advantages in decoding global spatial interaction principles.

This advancement provides a transformative analytical toolkit for investigating biological processes involving hierarchical spatial coordination, including cancer microenvironment evolution and organ developmental patterning. By introducing a field-theoretic model, this study pioneers paradigm shift in spatial transcriptomics analytics: transitioning from traditional localized discrete descriptions to continuous global system modeling, which significantly enhances spatial information utilization efficiency while establishing a novel framework for elucidating multi-scale spatial interaction networks.