Topology-Preserving Augmentation for CNN-Based Segmentation of Congenital Heart Defects from 3D Paediatric CMR

被引:1
作者
Byrne, Nick [1 ,2 ]
Clough, James R. [2 ]
Valverde, Isra [2 ]
Montana, Giovanni [3 ,4 ]
King, Andrew P. [2 ]
机构
[1] Guys & St Thomas NHS Fdn Trust, Med Phys, London, England
[2] Kings Coll London, Sch Biomed Engn & Imaging Sci, London, England
[3] Guys & St Thomas NHS Fdn Trust, Paediat Cardiol, London, England
[4] Univ Warwick, Warwick Mfg Grp, Coventry, W Midlands, England
来源
SMART ULTRASOUND IMAGING AND PERINATAL, PRETERM AND PAEDIATRIC IMAGE ANALYSIS, SUSI 2019, PIPPI 2019 | 2019年 / 11798卷
关键词
Image segmentation; Data augmentation; Topology;
D O I
10.1007/978-3-030-32875-7_20
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Patient-specific 3D printing of congenital heart anatomy demands an accurate segmentation of the thin tissue interfaces which characterise these diagnoses. Even when a label set has a high spatial overlap with the ground truth, inaccurate delineation of these interfaces can result in topological errors. These compromise the clinical utility of such models due to the anomalous appearance of defects. CNNs have achieved state-of-the-art performance in segmentation tasks. Whilst data augmentation has often played an important role, we show that conventional image resampling schemes used therein can introduce topological changes in the ground truth labelling of augmented samples. We present a novel pipeline to correct for these changes, using a fast-marching algorithm to enforce the topology of the ground truth labels within their augmented representations. In so doing, we invoke the idea of cardiac contiguous topology to describe an arbitrary combination of congenital heart defects and develop an associated, clinically meaningful metric to measure the topological correctness of segmentations. In a series of five-fold cross-validations, we demonstrate the performance gain produced by this pipeline and the relevance of topological considerations to the segmentation of congenital heart defects. We speculate as to the applicability of this approach to any segmentation task involving morphologically complex targets.
引用
收藏
页码:181 / 188
页数:8
相关论文
共 9 条
  • [1] Topology correction of segmented medical images using a fast marching algorith
    Bazin, Pierre-Louis
    Pham, Dzunq L.
    [J]. COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE, 2007, 88 (02) : 182 - 190
  • [2] Lequan Yu, 2017, Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017. 20th International Conference. Proceedings: LNCS 10434, P287, DOI 10.1007/978-3-319-66185-8_33
  • [3] Automatic Whole-Heart Segmentation in Congenital Heart Disease Using Deeply-Supervised 3D FCN
    Li, Jinpeng
    Zhang, Rongzhao
    Shi, Lin
    Wang, Defeng
    [J]. RECONSTRUCTION, SEGMENTATION, AND ANALYSIS OF MEDICAL IMAGES, 2017, 10129 : 111 - 118
  • [4] A GPU Based Diffusion Method for Whole-Heart and Great Vessel Segmentation
    Loesel, Philipp
    Heuveline, Vincent
    [J]. RECONSTRUCTION, SEGMENTATION, AND ANALYSIS OF MEDICAL IMAGES, 2017, 10129 : 121 - 128
  • [5] V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation
    Milletari, Fausto
    Navab, Nassir
    Ahmadi, Seyed-Ahmad
    [J]. PROCEEDINGS OF 2016 FOURTH INTERNATIONAL CONFERENCE ON 3D VISION (3DV), 2016, : 565 - 571
  • [6] Interactive Whole-Heart Segmentation in Congenital Heart Disease
    Pace, Danielle F.
    Dalca, Adrian V.
    Geva, Tal
    Powell, Andrew J.
    Moghari, Mehdi H.
    Golland, Polina
    [J]. MEDICAL IMAGE COMPUTING AND COMPUTER-ASSISTED INTERVENTION, PT III, 2015, 9351 : 80 - 88
  • [7] Dilated Convolutional Neural Networks for Cardiovascular MR Segmentation in Congenital Heart Disease
    Wolterink, Jelmer M.
    Leiner, Tim
    Viergever, Max A.
    Isgum, Ivana
    [J]. RECONSTRUCTION, SEGMENTATION, AND ANALYSIS OF MEDICAL IMAGES, 2017, 10129 : 95 - 102
  • [8] 3D FractalNet: Dense Volumetric Segmentation for Cardiovascular MRI Volumes
    Yu, Lequan
    Yang, Xin
    Qin, Jing
    Heng, Pheng-Ann
    [J]. RECONSTRUCTION, SEGMENTATION, AND ANALYSIS OF MEDICAL IMAGES, 2017, 10129 : 103 - 110
  • [9] Strengths and Pitfalls of Whole-Heart Atlas-Based Segmentation in Congenital Heart Disease Patients
    Zuluaga, Maria A.
    Biffi, Benedetta
    Taylor, Andrew M.
    Schievano, Silvia
    Vercauteren, Tom
    Ourselin, Sebastien
    [J]. RECONSTRUCTION, SEGMENTATION, AND ANALYSIS OF MEDICAL IMAGES, 2017, 10129 : 139 - 146