基于MRI影像组学的胶质瘤高低级别分级研究综述

Review of the Research on High-Level and Low-Level Grading of Glioma Based on MRI Radiomics

目的综述基于多模态MRI影像组学的胶质瘤分级研究进展,总结影响分级诊断准确性的因素。方法检索2012年1月至2021年12月PubMed、Scopus、Web of Science 3个数据库中与胶质瘤分级有关的影像组学的文献。根据纳入和排除标准,对文献进行筛选。获得文献中的敏感度、特异性、曲线下面积、模型特征筛选参数等,对筛选后的文献进行分析。结果通过检索,最终入组15篇文献。分析结果表明,使用多模态MRI进行等级分类的敏感度较高,逻辑回归模型和支持向量机模型应用较多,深度学习在胶质瘤分级中表现出色。结论影像组学方法在胶质瘤分级方面展现出了优异的结果,可为临床工作提供帮助。

Objective To summarize the research progress of glioma classification based on multi-modal MRI radiomics,and to summarize the factors affecting the accuracy of classification diagnosis.Methods From January 2012 to December 2021,PubMed,Scopus and Web of Science databases were searched for articles related to radiomics and glioma grading.The articles were screened according to inclusion and exclusion criteria.The sensitivity,specificity,area under curve and model feature screening parameters were obtained,and the screened articles were analyzed.Results Through retrieval,15 articles were finally included.The analysis results showed that the sensitivity of multi-modal MRI classification was higher,Logistic regression model and support vector machine model were widely used,and deep learning performed well in glioma classification.Conclusion The radiomics has showed excellent results in glioma grading,which can be helpful for clinical work.