Li Xiao, Ren Xuehan, Venugopal Raghavan
Computational Science and Informatics, Roche Diagnostics Solutions, Santa Clara, CA, USA.
Biostatistics, Clinical Data Science, Gilead Sciences, Inc., Foster City, CA, USA.
iScience. 2025 May 28;28(6):112765. doi: 10.1016/j.isci.2025.112765. eCollection 2025 Jun 20.
Entropy, a cornerstone of information theory, quantifies disorder, heterogeneity, and complexity in biological systems. This review explores its use in digital pathology and spatial omics, focusing on how entropy captures tissue architecture, spatial heterogeneity, and cellular organization. This review synthesizes entropy's theoretical foundations, spanning classical and spatially-aware metrics, and highlights its applications in mapping tissue heterogeneity, profiling microenvironments, and informing biomarker discovery through AI/ML. Key challenges include computational scalability, interpretability, and the need for standardization. We also discuss future directions, including graph-based entropy for cell networks, dynamic entropy for disease progression, and integrative approaches across molecular and spatial data. By merging theoretical precision with clinical applications, entropy-based methods offer promising tools for advancing precision medicine and personalized treatment strategies.
熵作为信息论的基石,量化了生物系统中的无序性、异质性和复杂性。本综述探讨了其在数字病理学和空间组学中的应用,重点关注熵如何捕捉组织结构、空间异质性和细胞组织。本综述综合了熵的理论基础,涵盖经典和空间感知指标,并强调了其在绘制组织异质性、分析微环境以及通过人工智能/机器学习为生物标志物发现提供信息方面的应用。关键挑战包括计算可扩展性、可解释性以及标准化的需求。我们还讨论了未来的方向,包括用于细胞网络的基于图的熵、用于疾病进展的动态熵以及跨分子和空间数据的综合方法。通过将理论精度与临床应用相结合,基于熵的方法为推进精准医学和个性化治疗策略提供了有前景的工具。