目的 分析基于多模态磁共振成像(Magnetic Resonance Imaging,MRI)影像组学鉴别胶质瘤及单发脑转移瘤的研究进展,得出提升鉴别准确性的要素.方法 通过检索PubMed、Web of Science及FMRS外文医学信息资源检索平台3个数据库,根据纳入排除标准,对纳入的文章提取数据来源、患者数量、MRI设备、MRI序列、肿瘤分割软件、分割方式、分割范围、分割类型、特征提取方法、筛选方法、机器学习分类器、最优的机器学习分类器等数据进行综合分析.结果 最终纳入12篇文献进行分析,大多数研究选择MRI传统结构序列,特征筛选方法选择最多的是最小绝对收缩和选择算子,使用最多且表现最佳的机器学习分类器为随机森林.结论 MRI影像组学方法在鉴别胶质瘤及单发脑转移瘤方面展现出了较高的准确性,为临床决策提高了较大帮助.
Application Research of MRI Radiomics in Differentiating Glioma from Single Brain Metastases
Objective To analyze the research progress in the identification of glioma and single brain metastases based on multimodal magnetic resonance imaging(MRI),and obtain the factors of improving accuracy in the identification.Methods Through searching three databases of PubMed,Web of Science and FMRS foreign medical information resource retrieval platform,according to the inclusion and exclusion criteria,a comprehensive analysis was made on data sources,number of patients,MRI equipment,MRI sequence,tumor segmentation software,segmentation methods,segmentation scopes,segmentation types,feature extraction methods,screening methods,machine learning classifiers and optimal machine learning classifiers of the included articles.Results A total of 12 articles were included for analysis.The traditional structural sequences of MRI were selected in most studies,least absolute shrinkage and selection operator was the most selected feature screening methods,and random forest was the machine learning classifier with the most use and the best performance.Conclusion MRI radiomics method shows high accuracy in differentiating glioma from single brain metastasis,which is of great help for clinical decision-making.
万方数据
维普
