基于U-Net的多图谱标签融合算法

芦玥; 马瑜; 王慧; 王原

doi:10.3788/YJYXS20193411.1091

您当前的位置：

首页 >

文章列表页 >

基于U-Net的多图谱标签融合算法

图像处理 | 更新时间：2020-07-05

- 基于U-Net的多图谱标签融合算法
- Multi-atlaslabel fusion based on U-Net
- 液晶与显示 2019年34卷第11期页码：1091-1103
- 作者机构：
  
  宁夏大学物理与电子电气工程学院, 宁夏银川 750021
- 作者简介：
  
  [ "芦玥(1995-), 女, 宁夏青铜峡人, 硕士研究生, 2017年于山东科技大学获得学士学位, 主要从事模式识别与图像处理方向的研究。E-mail:495400199@qq.com" ]
  [ "马瑜(1974-), 男, 宁夏西吉人, 工学博士, 教授, 硕士生导师, 2013年于西北工业大学获得博士学位, 主要从事计算机视觉与模式识别方向的研究。E-mail:mayu95@163.com" ]
- 基金信息：
  
  宁夏自然科学基金(NZ16009);宁夏高等学校科学研究项目(NGY2016015);2018年宁夏研究生教育教学改革研究与实践项目(YJG201811);宁夏大学研究生创新研究项目(GIP2019060);National Natural Science Foundation of Ningxia Autonomous Region(NZ16009);Ningxia Higher Education Research Project(NGY2016015);Ningxia Postgraduate Education Teaching Reform Research and Practice Project in 2018(YJG201811);Ningxia University Graduate Innovation Research Project(GIP2019060)
- DOI：10.3788/YJYXS20193411.1091
  中图分类号： TP391.4
- 收稿日期：2019-07-12，
  
  录用日期：2019-9-12，
  
  纸质出版日期：2019-11-05
- 稿件说明：
移动端阅览
芦玥, 马瑜, 王慧, 等. 基于U-Net的多图谱标签融合算法[J]. 液晶与显示, 2019,34(11):1091-1103.

Yue LU, Yu MA, Hui WANG, et al. Multi-atlaslabel fusion based on U-Net[J]. Chinese journal of liquid crystals and displays, 2019, 34(11): 1091-1103.
芦玥, 马瑜, 王慧, 等. 基于U-Net的多图谱标签融合算法[J]. 液晶与显示, 2019,34(11):1091-1103. DOI： 10.3788/YJYXS20193411.1091.

Yue LU, Yu MA, Hui WANG, et al. Multi-atlaslabel fusion based on U-Net[J]. Chinese journal of liquid crystals and displays, 2019, 34(11): 1091-1103. DOI： 10.3788/YJYXS20193411.1091.

摘要

为有效提高海马体多图谱分割算法的精度，将U-Net卷积神经网络用于海马体MR多图谱分割的标签融合。算法在图谱选择阶段，计算互信息、梯度相似性选择图谱，避免周围组织结构对图谱选择的干扰，选择与目标图谱更贴合的浮动图像组。在预处理阶段，提取以海马体为中心的感兴趣区域有效降低数据规模。在配准过程中，利用重采样代替粗配准环节，减少了“粗”配准环节所需时间，再采用具有良好的平滑性、连续性和拓扑保持性的微分同胚Demons精配准算法。在标签融合阶段，提出基于深度学习理论改进的U-Net多图谱MRI海马体分割算法。实验结果表明，改进的算法分割精度较于传统算法平均提高了5%，算法时间缩短了50%左右。改进后的基于U-Net的多图谱海马体分割算法对目标图像海马体的分割具有高精度、高效率的特点。

Abstract

In order to effectively improve the accuracy of multi-atlas segmentation algorithm of hippocampus

U-Net convolutional neural network is applied to label fusion of multi-atlas. The algorithm in the atlases selection performs mutual information and gradient similarity calculation

avoids the interference of the surrounding tissue structure on the atlas selection

selects the floating image group which is more suitable for the target map. In the pre-processing stage

extracting the region of interest centered on hippocampus can effectively reduce the size of data. In the registration process

re-sampling is used instead of the coarse registration

which reduces the time

and then uses the diffeomorphic demons algorithm

which has good smoothness

continuity and topological retentiveness. In the label fusion stage

an improved U-Net network based on deep learning theory for multi-atlas MRI hippocampal segmentation algorithm is proposed. The experimental results show that the segmentation accuracy of the improved algorithm is about 5% higher than that of the traditional algorithm

the algorithm time is reduced by about 50%. The improved U-Net network based multi-atlas hippocampal segmentation algorithm has the characteristics of high precision and high efficiency for segmentation of the hippocampus in target image.

关键词

Keywords

references

HOGAN R E, MARK K E, CHOUDHURI I, et al . Magnetic resonance imaging deformation-based segmentation of the hippocampus in patients with mesial temporal sclerosis and temporal lobe epilepsy[J]. Journal of Digital Imaging , 2000, 13(S1):217-218.

SHATTUCK D W, MIRZA M, ADISETIYO V, et al . Construction of a 3D probabilistic atlas of human cortical structures[J]. Neuroimage , 2008, 39(3):1064-1080.

JACK C R JR, BENTLEY M D, TWOMEY C K, et al . MR imaging-based volume measurements of the hippocampal formation and anterior temporal lobe:validation studies[J]. Radiology , 1990, 176(1):205-209.

KASS M, WITKIN A, TERZOPOULOS D. Snakes:active contour models[J]. International Journal of Computer Vision , 1988, 1(4):321-331.

ASHTON E A, BERG M J, PARKER K J, et al . Segmentation and feature extraction techniques, with applications to MRI head studies[J]. Magnetic Resonance in Medicine , 1995, 33(5):670-677.

KELEMEN A, SZEKELY G, GERIG G. Elastic model-based segmentation of 3-D neuroradiological data sets[J]. IEEE Transactions on Medical Imaging , 1999, 18(10):828-839.

CREMERS D. Dynamical statistical shape priors for level set-based tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence , 2006, 28(8):1262-1273.

ZHANG H Z, MORROW P, MCCLEAN S, et al . Contour detection of labelled cellular structures from serial ultrathin electron microscopy sections using GAC and prior analysis[C]// Proceedings of the 20081st Workshops on Image Processing Theory , Tools and Applications. Sousse: IEEE, 2008.

DAMBREVILLE S, RATHI Y, TANNENBAUM A. A framework for image segmentation using shape models and kernel space shape priors[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence , 2008, 30(8):1385-1399.

LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[C]// Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015.

RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]// Proceedings of the 18th International Conference on Medical Image Computing and Computer - Assisted Intervention . Munich: Springer, 2015: 234-241. https://link.springer.com/chapter/10.1007/978-3-319-24574-4_28

SHATTUCK D W, LEAHY R M.BrainSuite:an automated cortical surface identification tool[J]. Medical Image Analysis , 2002, 6(2):129-142.

ALJABAR P, HECKEMANN R A, HAMMERSA, et al . Multi-atlas based segmentation of brain images:atlas selection and its effect on accuracy[J]. Neuroimage , 2009, 46(3):726-738.

AWATE S P, ZHU(p)H, WHITAKER R T. How many templates does it take for a good segmentation?: error analysis in multiatlas segmentation as a function of database size[C]// Proceedings of the 2nd International Workshop on Multimodal Brain Image Analysis . Nice: Springer, 2012: 103-114. https://link.springer.com/chapter/10.1007/978-3-642-33530-3_9

夏瑞, 马瑜, 王文娜, 等.基于重采样改进的多图谱分割算法[J].计算机工程与设计, 2018, 39(8):2587-2592, 2659.

XIA R, MA Y, WANG W N, et al . Multi-atlas segmentation algorithm based on resampling[J]. Computer Engineering and Design , 2018, 39(8):2587-2592, 2659. (in Chinese)

VERCAUTEREN T, PENNEC X, PERCHANT A, et al . Diffeomorphic demons:efficient non-parametric image registration[J]. Neuroimage , 2009, 45(1 Suppl):S61-S72.

HUO J, WANG G H, WU Q M J, et al . Label fusion for multi-atlas segmentation based on majority voting[C]// Proceedings of the 12th International Conference Image Analysis and Recognition . Niagara Falls: Springer, 2015: 100-106. https://link.springer.com/chapter/10.1007%2F978-3-319-20801-5_11

WU G R, WANG Q, ZHANG D Q, et al . A generative probability model of joint label fusion for multi-atlas based brain segmentation[J]. Medical Image Analysis , 2014, 18(6):881-890.

WARFIELD S K, ZOU K H, WELLS W M. Simultaneous truth and performance level estimation (STAPLE):an algorithm for the validation of image segmentation[J]. IEEE Transactions on Medical Imaging , 2004, 23(7):903-921.

COUPÉ P, MANJÓN J V, FONOV V, et al . Patch-based segmentation using expert priors:application to hippocampus and ventricle segmentation[J]. NeuroImage , 2011, 54(2):940-954.

Brain-Development. Brain atlases[EB/OL]. http://biomedic.doc.ic.ac.uk/brain-development/ http://biomedic.doc.ic.ac.uk/brain-development/ .

Alzheimer's Disease Neuroimaging Initiative.[EB/OL]. http://adni.loni.usc.edu http://adni.loni.usc.edu .

JOHNSON H J, MCCORMICK M M, IBANEZ L. The ITK Software Guide Book 1:Introduction and Development Guidelines [M]. Kitware, Inc., 2015.

LIU H, YAN M, SONG E M, et al . Label fusion method based on sparse patch representation for the brain MRI image segmentation[J]. IET Image Processing , 2017, 11(7):502-511.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于ANTs配准的多图谱分割算法比较研究

基于改进U-Net网络的图像混合畸变校正方法

基于深度学习的磁瓦表面缺陷分割与识别