###
Journal of Software:2020.31(5):1573-1584

轮廓指导的层级混合多任务全卷积网络
何克磊,史颖欢,高阳
(计算机软件新技术国家重点实验室(南京大学), 江苏 南京 210023)
Contour-guided Hierarchically-fused Multi-task Fully Convolutional Network
HE Ke-Lei,SHI Ying-Huan,GAO Yang
(State Key Laboratory for Novel Software Technology (Nanjing University), Nanjing 210023, China)
Abstract
Chart / table
Reference
Similar Articles
Article :Browse 480   Download 439
Received:March 07, 2019    Revised:September 28, 2019
> 中文摘要: 传统的深度多任务网络通常在不同任务之间共享网络的大部分层(即特征表示层).由于这样做会忽视不同任务各自的特殊性,所以往往会制约其适应数据的能力.提出了一种层级混合的多任务全卷积网络HFFCN,以解决CT图像中的前列腺分割问题.特别地,使用一个多任务框架来解决这个问题.这个框架包括一个分割前列腺的主任务和一个回归前列腺边界的辅助任务.这里,第2个任务主要是用来精确地描述在CT图像中模糊的前列腺边界.因此,HFFCN架构是一个双分支的结构,包含一个编码主干和两个解码分支.不同于传统的多任务网络,提出了一个信息共享模块,用以在两个解码分支之间共享信息.这使得HFFCN可以学习任务的通用层级信息,同时保留一些不同任务各自的特征表示.在一个包含有313个病人的313张计划阶段图片的CT图像数据集上做了详细的实验.实验结果证明了HFFCN网络可以超越现有其他先进的分割方法或者传统的多任务学习模型.
Abstract:Conventional multi-task deep networks typically share most of the layers (i.e., layers for feature representations) across all tasks, which may limit their data fitting ability, as specificities of different tasks are inevitably ignored. This study proposes a hierarchically-fused multi-task fully-convolutional network, called HFFCN, to tackle the challenging task of prostate segmentation in CT images. Specifically, prostate segmentation is formulated into a multi-task learning framework, which includes a main task to segment prostate, and a supplementary task to regress prostate boundary. Here, the second task is applied to accurately delineating the boundary of the prostate, which is very unclear in CT images. Accordingly, the HFFCN uses a two-branch structure consisting of a shared encoding path and two complementary decoding paths. In contrast to the conventional multi-task networks, an information sharing (IS) module is also proposed to communicate at each level between the two decoding branches, by which the HFFCN endows the ability to learn hierarchically the complementary feature representations for different tasks, and also simultaneously preserve the specificities of learned feature representations for different tasks. The HFFCN is comprehensively evaluated on a large CT image dataset, including 313 images acquired from 313 patients. The experimental results demonstrate that the proposed HFFCN outperforms both the state-of-the-art segmentation methods and the conventional multi-task learning methods.
文章编号:     中图分类号:TP391    文献标志码:
基金项目:国家自然科学基金(61673203);江苏省重点研发计划专项基金(BE2018610) 国家自然科学基金(61673203);江苏省重点研发计划专项基金(BE2018610)
Foundation items:National Natural Science Foundation of China (61673203); Special Fund of Key Research and Development Program of Jiangsu Province (BE2018610)
Reference text:

何克磊,史颖欢,高阳.轮廓指导的层级混合多任务全卷积网络.软件学报,2020,31(5):1573-1584

HE Ke-Lei,SHI Ying-Huan,GAO Yang.Contour-guided Hierarchically-fused Multi-task Fully Convolutional Network.Journal of Software,2020,31(5):1573-1584