陈皓^{1, 2,,},
李广¹,
刘洋¹,
强永乾³
1.西安邮电大学计算机学院　西安　710121
2.陕西省网络数据分析与智能处理重点实验室　西安　710121
3.西安交通大学第一附属医院　西安　710061
基金项目:国家自然科学基金(61876138, 61203311)，陕西省自然科学基金(2019JM-365)，陕西省教育厅自然科学专项(17JK0701)，西安邮电大学研究生创新基金(CXJJ2017036)

详细信息

作者简介:陈皓：男，1978年生，博士，副教授，硕士研究生导师，主要研究方向为医疗大数据
李广：男，1995年生，硕士生，研究方向为计算智能与数据挖掘
刘洋：男，1995年生，硕士生，研究方向为计算智能与数据挖掘
强永乾：男，1965年生，博士，副教授，硕士研究生导师，研究方向为医学影像学

通讯作者:陈皓　chenhao@xupt.edu.cn

中图分类号:TP391.41, R445.2

计量

文章访问数:1136
HTML全文浏览量:425
PDF下载量:93
被引次数:0

出版历程

收稿日期:2020-01-09
修回日期:2020-06-15
网络出版日期:2020-07-22
刊出日期:2021-04-20

A Glioma Detection and Segmentation Method in MR Imaging

Hao CHEN^{1, 2,,},
Guang LI¹,
Yang LIU¹,
Yongqian QIANG³
1. School of Computer, Xi’an University of Posts & Telecommunications, Xi’an 710121, China
2. Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing, Xi’an University of Post and Telecommunications, Xi’an 710121, China
3. First Affiliated Hospital of Xi’an Jiaotong University, Xi’an 710061, China
Funds:The National Natural Science Foundation of China (61876138, 61203311), The Natural Science Basic Research Program of Shaanxi Province (2019JM-365), The Scientific Research Program Funded by Shaanxi Provincial Education Department of China (17JK0701), The Graduate Innovation Foundation of Xi’an University of Posts & Telecommunications (CXJJ2017036)


摘要
摘要:针对磁共振图像(MRI)进行脑胶质瘤检测及病灶分割对临床治疗方案的选择和手术实施过程的引导都有着重要的价值。为了提高脑胶质瘤的检测效率和分割准确率，该文提出了一种两阶段计算方法。首先，设计了一个轻量级的卷积神经网络，并通过该网络完成MR图像中肿瘤的快速检测及大致定位；接着，通过集成学习过程对肿瘤周围水肿、肿瘤非增强区、肿瘤增强区和正常脑组织等4种不同区域进行分类与彼此边界的精细分割。为提高分割的准确率，在MR图像中提取了416维影像组学特征并与128维通过卷积神经网络提取的高阶特征进行组合和特征约简，将特征约简后产生的298维特征向量用于分类学习。为对算法的性能进行验证，在BraTS2017数据集上进行了实验，实验结果显示该文提出的方法能够快速检测并定位肿瘤，同时相比其它方法，整体分割精度也有明显提升。
关键词:肿瘤检测/
病灶边界分割/
特征选择/
集成学习
Abstract:The glioma detection and focus segmentation in Magnetic Resonance Imaging (MRI) has important value for the therapeutic schedule selection and the surgical operations. In order to improve the detection efficiency and segmentation accuracy for glioma, this paper proposes a two-stage calculating method. First, a light convolutional neural network is designed to implement rapidly detection and localization for the glioma in MR images. Then, the peritumoral edema, non-enhancing tumor, enhancing tumor, and normal are classified and segmented from each other through an Ensemble Learning (EL) process. In order to improve the accuracy of segmentation, 416 radiomics features extracted from multi-modal MR images and 128 CNN features extracted by a convolutional neural network are mixed. The feature vector consisting of 298 features for classification learning are formed after a feature reduction process. In order to verify the performance of the proposed algorithm, experiments are carried out on the BraTS2017 dataset. The experimental results show that the proposed method can quickly detect and locate the tumor. The overall segmentation accuracy is improved distinctly with respect to 4 state-of-the-art approaches.
Key words:Tumor detection/
Focus segmentation/
Features selection/
Ensemble Learning (EL)



PDF全文下载地址:

https://jeit.ac.cn/article/exportPdf?id=ca696188-8e83-458a-a173-15e4066d0057