Location Sensitive Deep Convolutional Neural Networks for Segmentation of White Matter Hyperintensities

被引:182
作者
Ghafoorian, Mohsen [1 ,2 ]
Karssemeijer, Nico [2 ]
Heskes, Tom [1 ]
van Uden, Inge W. M. [3 ]
Sanchez, Clara I. [2 ]
Litjens, Geert [2 ]
de Leeuw, Frank-Erik [3 ]
van Ginneken, Bram [2 ]
Marchiori, Elena [1 ]
Platel, Bram [2 ]
机构
[1] Radboud Univ Nijmegen, Inst Comp & Informat Sci, Nijmegen, Netherlands
[2] Radboud Univ Nijmegen, Dept Radiol & Nucl Med, Diagnost Image Anal Grp, Med Ctr, Nijmegen, Netherlands
[3] Radboud Univ Nijmegen, Donders Inst Brain Cognit & Behav, Dept Neurol, Med Ctr, Nijmegen, Netherlands
关键词
MULTIPLE-SCLEROSIS LESIONS; AUTOMATIC SEGMENTATION; MR-IMAGES; ALZHEIMERS-DISEASE; MODEL; QUANTIFICATION; VOLUME; PEOPLE; IMPACT;
D O I
10.1038/s41598-017-05300-5
中图分类号
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号
070301 [无机化学]; 070403 [天体物理学]; 070507 [自然资源与国土空间规划学]; 090105 [作物生产系统与生态工程];
摘要
The anatomical location of imaging features is of crucial importance for accurate diagnosis in many medical tasks. Convolutional neural networks (CNN) have had huge successes in computer vision, but they lack the natural ability to incorporate the anatomical location in their decision making process, hindering success in some medical image analysis tasks. In this paper, to integrate the anatomical location information into the network, we propose several deep CNN architectures that consider multiscale patches or take explicit location features while training. We apply and compare the proposed architectures for segmentation of white matter hyperintensities in brain MR images on a large dataset. As a result, we observe that the CNNs that incorporate location information substantially outperform a conventional segmentation method with handcrafted features as well as CNNs that do not integrate location information. On a test set of 50 scans, the best configuration of our networks obtained a Dice score of 0.792, compared to 0.805 for an independent human observer. Performance levels of the machine and the independent human observer were not statistically significantly different (p-value = 0.06).
引用
收藏
页数:12
相关论文
共 93 条
[1]
Fully automatic segmentation of white matter hyperintensities in MR images of the elderly [J].
Admiraal-Behloul, F ;
van den Heuvel, DMJ ;
Olofsen, H ;
van Osch, MJP ;
van der Grond, J ;
van Buchem, MA ;
Relber, JHC .
NEUROIMAGE, 2005, 28 (03) :607-617
[2]
Probabilistic segmentation of white lesions in MR imaging [J].
Anbeek, P ;
Vincken, KL ;
van Osch, MJP ;
Bisschops, RHC ;
van der Grond, J .
NEUROIMAGE, 2004, 21 (03) :1037-1044
[3]
[Anonymous], ARXIV160107014
[4]
[Anonymous], ARXIV13090261
[5]
[Anonymous], 2016, ARXIV160202130
[6]
[Anonymous], 2016, ARXIV160705194
[7]
[Anonymous], 2017, ARXIV170205747
[8]
[Anonymous], 2016, VOXRESNET DEEP VOXEL
[9]
[Anonymous], 2013, PROC 30 INT C MACH L
[10]
[Anonymous], SPIE MED IMAGING