基于CT影像组学的多发性骨髓瘤骨骼 生境异质性分析

CT Radiomics-Based Analysis of Skeletal Habitat Heterogeneity in Multiple Myeloma

摘要: 多发性骨髓瘤(MM)骨骼浸润的骨髓微环境空间异质性是临床诊断与疗效评估的关键难点，传统检测技术难以全面反映其病理特征。本文以CT影像组学为核心技术切入点，系统阐述MM骨骼生境异质性的病理生理基础，包括骨髓微环境的区域异质性特征、骨髓脂肪组织(BMAT)与肿瘤细胞的动态互作及微血管密度(MVD)的空间分布促瘤机制；梳理CT影像组学在多参数特征提取、机器学习骨骼分割中的技术进展，剖析肋骨、胸骨柄、胸椎等不同骨骼部位的特异性影像组学特征；明确基于灰度共生矩阵(GLCM)的细胞浸润度预测、纹理特征与骨髓纤维化的相关性等核心影像生物标志物的临床价值；总结该技术在MM早期浸润检测、治疗反应定量化评估、预后分层模型构建中的转化应用成果。同时分析当前研究面临的多模态数据融合算法优化、区域特异性生物标志物标准化等挑战，并展望未来研究方向。本文指出，CT影像组学实现了MM骨髓微环境异质性的无创定量评估，突破了传统骨髓穿刺的局限性，为MM精准诊疗提供了全新的技术手段和科学依据，其对骨髓瘤–骨骼生境互作规律的揭示也为肿瘤生态学研究提供了重要范式参考。

Abstract: The spatial heterogeneity of the bone marrow microenvironment in multiple myeloma (MM) with skeletal infiltration represents a major challenge in clinical diagnosis and treatment evaluation, as conventional detection techniques fail to comprehensively capture its pathological characteristics. This review focuses on CT radiomics as a core technological approach, systematically elucidating the pathophysiological basis of skeletal habitat heterogeneity in MM, including regional heterogeneity within the bone marrow microenvironment, dynamic interactions between bone marrow adipose tissue (BMAT) and tumor cells, and the tumor-promoting mechanisms associated with the spatial distribution of microvascular density (MVD). It outlines advances in CT radiomics for multiparametric feature extraction and machine learning-based skeletal segmentation, and analyzes site-specific radiomic features across different skeletal regions, including the ribs, sternum, and thoracic vertebrae. Key imaging biomarkers with clinical relevance are identified, such as the prediction of cellular infiltration using gray-level co-occurrence matrix (GLCM) features and the correlation between textural features and bone marrow fibrosis. The review further summarizes translational applications of this technology in early detection of MM infiltration, quantitative assessment of treatment response, and development of prognostic stratification models. Current challenges, including optimization of multimodal data fusion algorithms and standardization of region-specific biomarkers, are discussed alongside future research directions. This review indicates that CT radiomics enables non-invasive quantitative assessment of bone marrow microenvironment heterogeneity in MM, overcoming the limitations of conventional bone marrow biopsy. It provides a novel technological framework and scientific basis for precision diagnosis and treatment of MM, while also offering an important paradigm for tumor ecology research through its insights into the dynamics of myeloma-bone habitat interactions.