Choroid segmentation from Optical Coherence Tomography with graph edge weights learned from deep convolutional neural networks

被引:110
作者
Sui, Xiaodan [1 ,2 ,3 ]
Zheng, Yuanjie [1 ,2 ,3 ]
Wei, Benzheng [5 ]
Bi, Hongsheng [6 ,7 ]
Wu, Jianfeng [5 ]
Pan, Xuemei [7 ]
Yin, Yilong [8 ]
Zhang, Shaoting [4 ]
机构
[1] Shandong Normal Univ, Sch Informat Sci & Engn, Jinan, Shandong, Peoples R China
[2] Shandong Normal Univ, Inst Life Sci, Jinan, Shandong, Peoples R China
[3] Shandong Normal Univ, Key Lab Intelligent Informat Proc, Jinan, Shandong, Peoples R China
[4] Univ N Carolina, Dept Comp Sci, Charlotte, NC 28223 USA
[5] Shandong Univ Tradit Chinese Med, Jinan, Shandong, Peoples R China
[6] Shandong Univ TCM, Inst Eye, Jinan, Shandong, Peoples R China
[7] Shandong Univ TCM, Affiliated Eye Hosp, Jinan, Shandong, Peoples R China
[8] Shandong Univ, Sch Comp Sci & Technol, Jinan, Shandong, Peoples R China
关键词
Image segmentation; Learning; CNN; Choroid; OCT; AUTOMATIC SEGMENTATION; MACULAR DEGENERATION; THICKNESS; HEALTHY; IMAGES; OCT;
D O I
10.1016/j.neucom.2017.01.023
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Examining choroid in Optical Coherence Tomography (OCT) plays a vital role in pathophysiologic factors of many ocular conditions. Among the existing approaches to detecting choroidal boundaries, graph-searching based techniques belong to the state-of-the-art. However, most of these techniques rely on hand-crafted models on the graph-edge weight and their performances are limited mainly due to the weak choroidal boundaries, textural structure of the choroid, inhomogeneity of the textural structure of the choroid and great variation of the choroidal thickness. In order to circumvent this limitation, we present a multi-scale and end-to-end convolutional network architecture where an optimal graph-edge weight can be learned directly from raw pixels. Our method operates on multiple scales and combines local and global information from the 2D OCT image. Experimental results obtained based on 912 OCT B-scans show that our learned graph-edge weights outperform conventional hand-crafted ones and behave robustly and accurately no matter the OCT image is from normal subjects or patients for whom significant retinal structure variations can be observed.
引用
收藏
页码:332 / 341
页数:10
相关论文
共 43 条
  • [1] Automatic segmentation of choroidal thickness in optical coherence tomography
    Alonso-Caneiro, David
    Read, Scott A.
    Collins, Michael J.
    [J]. BIOMEDICAL OPTICS EXPRESS, 2013, 4 (12): : 2795 - 2812
  • [2] [Anonymous], ARXIV160508401
  • [3] BENNETT G, 1955, Br J Ophthalmol, V39, P605, DOI 10.1136/bjo.39.10.605
  • [4] Bertasius G, 2015, PROC CVPR IEEE, P4380, DOI 10.1109/CVPR.2015.7299067
  • [5] Analysis of Choroidal Morphologic Features and Vasculature in Healthy Eyes Using Spectral-Domain Optical Coherence Tomography
    Branchini, Lauren A.
    Adhi, Mehreen
    Regatieri, Caio V.
    Nandakumar, Namrata
    Liu, Jonathan J.
    Laver, Nora
    Fujimoto, James G.
    Duker, Jay S.
    [J]. OPHTHALMOLOGY, 2013, 120 (09) : 1901 - 1908
  • [6] Bron AJ., 1997, WOLFFS ANATOMY EYE O, V8th
  • [7] Automated choroid segmentation based on gradual intensity distance in HD-OCT images
    Chen, Qiang
    Fan, Wen
    Niu, Sijie
    Shi, Jiajia
    Shen, Honglie
    Yuan, Songtao
    [J]. OPTICS EXPRESS, 2015, 23 (07): : 8974 - 8994
  • [8] Automatic segmentation of seven retinal layers in SDOCT images congruent with expert manual segmentation
    Chiu, Stephanie J.
    Li, Xiao T.
    Nicholas, Peter
    Toth, Cynthia A.
    Izatt, Joseph A.
    Farsiu, Sina
    [J]. OPTICS EXPRESS, 2010, 18 (18): : 19413 - 19428
  • [9] Ciresan D., 2012, Advances in Neural Information Processing Systems 25 (NIPS 2012), P1
  • [10] Segmentation of Choroidal Boundary in Enhanced Depth Imaging OCTs Using a Multiresolution Texture Based Modeling in Graph Cuts
    Danesh, Hajar
    Kafieh, Raheleh
    Rabbani, Hossein
    Hajizadeh, Fedra
    [J]. COMPUTATIONAL AND MATHEMATICAL METHODS IN MEDICINE, 2014, 2014