基于机器学习的脑胶质瘤多模态影像分析  

A research on multi-modal MRI analysis based on machine learning for brain glioma

作　　者：吴强[1,2] 何泽鲲 刘琚[1,2] 崔晓萌 孙双 石伟 WU Qiang;HE Zekun;LIU Ju;CUI Xiaomeng;SUN Shuang;SHI Wei(School of Information Science and Engineering,Shandong University,Qingdao 266237,Shandong,China;Institute of Brain and Brain-Inspired Science,Shandong University,Jinan 250012,Shandong,China)

机构地区：[1]山东大学信息科学与工程学院,山东青岛266237 [2]山东大学脑与类脑科学研究院,山东济南250012

出　　处：《山东大学学报（医学版）》2020年第8期81-87,共7页Journal of Shandong University:Health Sciences

基　　金：山东大学基本科研业务费专项资金(2017JC013);山东省重大创新工程(2017CXGC1504)。

摘　　要：脑部胶质瘤是临床中常见的一种原发性脑肿瘤,具有复发率高、死亡率高以及治愈率低的特点。常规临床诊断主要依靠计算机断层扫描(CT)和磁共振成像(MRI)检查技术进行鉴别。随着成像技术和机器学习方法的不断发展,多模态影像智能分析技术已经逐步成为研究热点,在脑胶质瘤的病灶分割测量、肿瘤分级、预后生存周期预测和基因型辨别等方面具有重要的应用前景。本文重点介绍基于机器学习和多模态影像在脑胶质瘤临床辅助诊断和预后评估中的应用进展。Brain glioma,a common primary brain tumor,has characteristics of high recurrence rate,high death rate and low cure rate.Conventional clinical diagnosis mainly depends on CT and MRI.With the development of imaging technology and machine learning methods,multi-modal image intelligent analysis technology has gradually become a research hotspot,which has an important application prospect in brain glioma lesion segmentation and measurement,tumor classification,overall survival prediction and genotype identification.This paper updates the application of machine learning and multi-modal imaging in the clinical diagnosis and prognosis of brain glioma.

关 键 词：脑部胶质瘤 机器学习 多模态磁共振影像 影像病灶分割 生存周期预测 基因型预测 

分 类 号：R574[医药卫生—消化系统]