期刊
MEDICAL IMAGE ANALYSIS
卷 36, 期 -, 页码 61-78出版社
ELSEVIER
DOI: 10.1016/j.media.2016.10.004
关键词
3D convolutional neural network; Fully connected CRF; Segmentation Brain lesions; Deep learning
类别
资金
- EPSRC [EP/N023668/1]
- European Commission
- Medical Research Council (UK) [G9439390 ID 65883]
- UK National Institute of Health Research Biomedical Research Centre at Cambridge
- UK Department of Health
- Imperial College London
- Health Foundation/Academy of Medical Sciences Clinician Scientist Fellowship
- NIHR Senior Investigator Award
- NVIDIA Corporation
- Academy of Medical Sciences (AMS) [AMS-CSF4-Newcombe] Funding Source: researchfish
- Engineering and Physical Sciences Research Council [EP/N023668/1] Funding Source: researchfish
- Medical Research Council [G1000183B, G9439390] Funding Source: researchfish
- National Institute for Health Research [NF-SI-0512-10090, ACF-2009-14-007] Funding Source: researchfish
- EPSRC [EP/N023668/1] Funding Source: UKRI
- MRC [G9439390] Funding Source: UKRI
We propose a dual pathway, 11-layers deep, three-dimensional Convolutional Neural Network for the challenging task of brain lesion segmentation. The devised architecture is the result of an in-depth analysis of the limitations of current networks proposed for similar applications. To overcome the computational burden of processing 3D medical scans, we have devised an efficient and effective dense training scheme which joins the processing of adjacent image patches into one pass through the network while automatically adapting to the inherent class imbalance present in the data. Further, we analyze the development of deeper, thus more discriminative 3D CNNs. In order to incorporate both local and larger contextual information, we employ a dual pathway architecture that processes the input images at multiple scales simultaneously. For post-processing of the network's soft segmentation, we use a 3D fully connected Conditional Random Field which effectively removes false positives. Our pipeline is extensively evaluated on three challenging tasks of lesion segmentation in multi-channel MRI patient data with traumatic brain injuries, brain tumours, and ischemic stroke. We improve on the state-of-the-art for all three applications, with top ranking performance on the public benchmarks BRATS 2015 and ISLES 2015. Our method is computationally efficient, which allows its adoption in a variety of research and clinical settings. The source code of our implementation is made publicly available. (C) 2016 The Authors. Published by Elsevier B.V.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据