Please wait a minute...
浙江大学学报(工学版)  2019, Vol. 53 Issue (2): 355-363    DOI: 10.3785/j.issn.1008-973X.2019.02.019
计算机与控制工程     
胶质母细胞瘤多模态磁共振图像自动分割
赖小波1(),张学群2,许茂盛3
1. 浙江中医药大学 医学技术学院,浙江 杭州 310053
2. 浙江大学 流体传动与控制国家重点实验室,浙江 杭州 310027
3. 浙江中医药大学 第一临床医学院,浙江 杭州 310006
Automated segmentation for multi-modal magnetic resonance image of glioblastoma multiforme
Xiao-bo LAI1(),Xue-qun ZHANG2,Mao-sheng XU3
1. Medical Technology College, Zhejiang Chinese Medical University, Hangzhou 310053, China
2. State Key Laboratory of Fluid Power Transmission and Control, Zhejiang University, Hangzhou 310027, China
3. First Clinical Medical College, Zhejiang Chinese Medical University, Hangzhou 310006, China
 全文: PDF(1823 KB)   HTML
摘要:

针对大多数传统胶质母细胞瘤(GBM)多模态磁共振(MR)图像分割算法未能将整个肿瘤细分为不同子区域的问题,提出基于混合特征和先验知识的GBM多模态MR图像自动分割算法. 配准GBM多模态MR图像,将头部区域方位调整到近似未旋转位置,并利用N4ITK法进行偏置场校正. 在提取GBM多模态MR图像局部位置特征、强度特征、纹理特征、对称特征和上下文特征后,应用随机森林分类器初步分割GBM多模态MR图像. 考虑GBM肿瘤解剖结构先验知识,移除小区域和中值滤波后得到最终分割结果. 以Dice相似性系数作为评价指标,利用所提出的算法对TCGA-GBM和CH-GBM数据库中整个肿瘤进行分割,获得的平均Dice相似性系数分别为0.871、0.882. 结果表明,该算法能以较高的准确率分割GBM多模态MR图像,适用于临床GBM多模态MR图像分割任务.

关键词: 胶质母细胞瘤(GBM)多模态磁共振(MR)图像自动分割混合特征先验知识    
Abstract:

A glioblastoma multiforme (GBM) multi-modal magnetic resonance (MR) image automated segmentation algorithm based on hybrid features and prior knowledge was proposed, as most traditional GBM multi-modal MR image segmentation algorithms failed to subdivide the whole tumor into different sub-regions. The head region was adjusted to the approximate unrotated position once the GBM multi-modal MR image was registered, and the bias field correction was performed by the N4ITK method. A random forest classifier was applied to initially segment GBM multi-modal MR image after the extraction of the local location features, intensity features, texture features, symmetric features and contextual features of GBM multi-modal MR image. The final segmentation results were obtained by removing small regions and median filtering, based on the prior knowledge of the anatomical structure of GBM tumor. The Dice similarity coefficient was adopted as an evaluation metric, and the average Dice similarity coefficient values were 0.871 and 0.882 for segmenting the whole tumor in TCGA-GBM and CH-GBM databases by the proposed algorithm, respectively. Results indicated that the proposed method is suitable for clinical application of GBM multi-modal MR image segmentation task with relative high accuracy.

Key words: glioblastoma multiforme (GBM)    multi-modal magnetic resonance (MR) image    automatic segmentation    hybrid feature    prior knowledge
收稿日期: 2018-04-03 出版日期: 2019-02-21
CLC:  TP 751  
作者简介: 赖小波(1981—),男,副教授,硕导,博士,从事数字医学影像处理与分析研究. orcid.org/0000-0002-6543-8316. E-mail: dmia_lab@zcmu.edu.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章  
赖小波
张学群
许茂盛

引用本文:

赖小波,张学群,许茂盛. 胶质母细胞瘤多模态磁共振图像自动分割[J]. 浙江大学学报(工学版), 2019, 53(2): 355-363.

Xiao-bo LAI,Xue-qun ZHANG,Mao-sheng XU. Automated segmentation for multi-modal magnetic resonance image of glioblastoma multiforme. Journal of ZheJiang University (Engineering Science), 2019, 53(2): 355-363.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2019.02.019        http://www.zjujournals.com/eng/CN/Y2019/V53/I2/355

图 1  GBM多模态MR图像自动分割框架示意图
图 2  N4ITK偏置场校正前、后图像
图 3  FLAIR切片图像经LM滤波器组滤波处理后得到的48个响应
图 4  FLAIR模态磁共振图像体素的对称特征提取示意图
图 5  坏死区体素的上下文特征提取示意图
参数 数值
切片大小/像素 512×512
切片间距/mm T1WI-Pre:3.0~6.5,T1WI-Post:
2.5~6.5,FLAIR:2.5~6.5
像素间距/mm T1WI-Pre:0.429 7~0.938 0,T1WI-Post:
0.429 7~0.940 0,FLAIR:0.429 7~0.938 0
重复时间/ms T1WI-Pre:416.664 0~3 379.600 0,T1WI-Post:
4.944 3~285.600,FLAIR:8 002~11 000
回波/ms T1WI-Pre:6.356 0~15.000 0,T1WI-Post:
2.1~15.0,FLAIR:120.3~155.0
层面厚度/mm T1WI-Pre:3~5,T1WI-Post:
1.4~5.0,FLAIR:2.5~5.0
表 1  TCGA-GBM和CH-GBM数据库的详细信息
图 6  TCGA-GBM数据库自动分割结果与手动分割结果对比
图 7  CH-GBM数据库自动分割结果与手动分割结果对比
分区 Ddic Ssen Sspe
TCGA-GBM CH-GBM TCGA-GBM-CH TCGA-GBM CH-GBM TCGA-GBM-CH TCGA-GBM CH-GBM TCGA-GBM-CH
C 0.871 0.882 0.877 0.852 0.867 0.858 0.994 8 0.998 7 0.998 4
C1 0.863 0.872 0.866 0.875 0.889 0.881 0.995 9 0.999 2 0.997 5
C2 0.759 0.776 0.761 0.741 0.758 0.747 0.996 1 0.996 9 0.996 5
C3 0.775 0.783 0.781 0.748 0.773 0.764 0.994 1 0.995 7 0.995 2
C4 0.686 0.692 0.689 0.674 0.682 0.679 0.993 7 0.995 6 0.994 3
表 2  所提算法的Dice相似性系数、敏感度、特异性性能评估
1 PRASANNA P, PATEL J, PARTOVI S, et al Radiomic features from the peritumoral brain parenchyma on treatment-naive multi-parametric MR imaging predict long versus short-term survival in glioblastoma multiforme: preliminary findings[J]. European Radiology, 2017, 27 (10): 4198- 4199
doi: 10.1007/s00330-017-4815-y
2 RAVIKANTH R Advanced magnetic resonance imaging of glioblastoma multiforme[J]. Journal of Neurosciences in Rural Practice, 2017, 8 (3): 439- 440
doi: 10.4103/jnrp.jnrp_423_16
3 ZACHARAKI E, MORITA N, BHATT P, et al Survival analysis of patients with high-grade gliomas based on data mining of imaging variables[J]. American Journal of Neuroradiology, 2012, 33 (6): 1065- 1071
doi: 10.3174/ajnr.A2939
4 CLARK M C, HALL L O, GOLDGOF D B, et al Automatic tumor segmentation using knowledge-based techniques[J]. IEEE Transactions on Medical Imaging, 1998, 17 (2): 187- 201
doi: 10.1109/42.700731
5 FLETCHER-HEATH L M, HALL L O, GOLDGOF D B, et al Automatic segmentation of non-enhancing brain tumors in magnetic resonance images[J]. Artificial Intelligence in Medicine, 2001, 21 (1?3): 43- 63
doi: 10.1016/S0933-3657(00)00073-7
6 SHI J B, MALIK J Normalized cuts and image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22 (8): 888- 905
doi: 10.1109/34.868688
7 CORSO J J, SHARON E, DUBE S, et al Efficient multilevel brain tumor segmentation with integrated bayesian model classification[J]. IEEE Transactions on Medical Imaging, 2008, 27 (5): 629- 640
doi: 10.1109/TMI.2007.912817
8 CHINNADURAI V, CHANDRASHEKHAR G D Neuro-levelset system based segmentation in dynamic susceptibility contrast enhanced and diffusion weighted magnetic resonance images[J]. Pattern Recognition, 2012, 45 (9): 3501- 3511
doi: 10.1016/j.patcog.2012.02.038
9 POPURI K, COBZAS D, MURTHA A, et al 3D variational brain tumor segmentation using Dirichlet priors on a clustered feature set[J]. International Journal of Computer Assisted Radiology and Surgery, 2012, 7 (4): 493- 506
doi: 10.1007/s11548-011-0649-2
10 POPE W B, SAYRE J, PERLINA A, et al MR imaging correlates of survival in patients with high-grade gliomas[J]. American Journal of Neuroradiology, 2005, 26 (10): 2466- 2474
11 GREVE D N, FISCHL B Accurate and robust brain image alignment using boundary-based registration[J]. Neuroimage, 2009, 48 (1): 63- 72
doi: 10.1016/j.neuroimage.2009.06.060
12 ALEXANDER V T, OLIVIER C, ISABELLE B Evaluation of the symmetry plane in 3D MR brain images[J]. Pattern Recognition Letters, 2003, 24 (14): 2219- 2233
doi: 10.1016/S0167-8655(03)00049-7
13 TUSTISON N J, AVANTS B B, COOK P A, et al N4ITK: improved N3 bias correction[J]. IEEE Transactions on Medical Imaging, 2010, 29 (6): 1310- 1320
doi: 10.1109/TMI.2010.2046908
14 VARMA M, ZISSERMAN A A statistical approach to texture classification from single images[J]. International Journal of Computer Vision, 2005, 62 (1): 61- 81
15 TUZIKOV A V, COLLIOT O, BLOCH I Evaluation of the symmetry plane in 3D MR brain images[J]. Pattern Recognition Letters, 2003, 24 (14): 2219- 2233
16 BAUER S, NOLTE L P, REYES M Fully automatic segmentation of brain tumor images using support vector machine classification in combination with hierarchical conditional random field regularization[J]. Medical Image Computing and Computer-Assisted Intervention, 2011, 14: 354- 361
17 GOOYA A, POHL K, BILELLO M, et al GLISTR: glioma image segmentation and registration[J]. IEEE Transactions on Medical Imaging, 2012, 31 (10): 1941- 1954
doi: 10.1109/TMI.2012.2210558
[1] 高雪艳,潘安宁,杨扬. 基于图像混合特征的城市绿地遥感图像配准[J]. 浙江大学学报(工学版), 2019, 53(6): 1205-1217.