从白质高信号到认知障碍：脑小血管病影像组学研究的现状与展望

Bridging White Matter Hyperintensities and Cognitive Dysfunction: The Status and Prospect of Radiomics in Cerebral Small Vessel Disease

摘要: 脑小血管病(Cerebral small vessel disease, CSVD)是老年人血管性认知障碍及混合性痴呆的重要病因，其影像学标志性表现——白质高信号(white matter hyperintensity, WMH)与患者执行功能、信息处理速度等认知领域损害密切相关。近年来，影像组学(Radiomics)作为一种基于高通量特征提取的影像分析方法，通过挖掘常规医学图像中人眼难以辨识的定量特征，为CSVD相关认知障碍的客观评估与机制研究提供了新的技术路径。本文系统综述了影像组学在该领域的研究进展，重点探讨其在WMH异质性量化、认知功能预测及病理机制解析三方面的应用价值。研究表明，影像组学结合机器学习与多模态影像融合策略，能够有效克服传统影像评估中存在的主观性强、可重复性不足等局限，显著提升对CSVD患者认知状态的早期识别与精准评估能力。未来，通过建立标准化的数据处理流程、增强模型可解释性，影像组学有望在揭示血管性与神经退行性认知障碍的共同病理机制方面发挥关键作用，推动精准神经科学的发展。

Abstract: Cerebral small vessel disease (CSVD) is a key cause of vascular cognitive impairment and mixed dementia in the elderly. Its characteristic radiological expression, white matter hyperintensity (WMH), is directly linked to cognitive impairments in areas such as executive function and information processing speed. In recent years, radiomics—an image analysis method based on high-throughput feature extraction—has developed as a new technical pathway for objectively assessing and understanding CSVD-related cognitive impairment and its causes. This method accomplishes this by extracting quantitative information from traditional medical photographs that are difficult for the human eye to detect. This comprehensive review investigates radiomics research achievements in this field, with an emphasis on their utility in assessing WMH heterogeneity, predicting cognitive performance, and understanding pathogenic pathways. According to studies, combining radiomics with machine learning and multimodal image fusion strategies effectively overcomes the limitations inherent in traditional imaging assessments, such as high subjectivity and low reproducibility, significantly improving the early detection and precise evaluation of cognitive status in CSVD patients. Moving forward, radiomics has the potential to elucidate the overlapping pathogenic pathways underlying vascular and neurodegenerative cognitive deficits, boosting precision neuroscience.