TY - JOUR
T1 - Cascaded MultiTask 3-D Fully Convolutional Networks for Pancreas Segmentation
AU - Xue, Jie
AU - He, Kelei
AU - Nie, Dong
AU - Adeli, Ehsan
AU - Shi, Zhenshan
AU - Lee, Seong Whan
AU - Zheng, Yuanjie
AU - Liu, Xiyu
AU - Li, Dengwang
AU - Shen, Dinggang
N1 - Funding Information:
Manuscript received October 25, 2018; revised March 22, 2019 and October 7, 2019; accepted November 18, 2019. Date of publication December 18, 2019; date of current version March 17, 2021. The work of J. Xue was supported in part by the National Natural Science Foundation of China (NSFC) under Grant 61802234 and Grant 61640201, in part by the China Post-Doctoral Project under Grant 2017M612339, and in part by the China Scholarship Council under Grant 201708370073. The work of Y. Zheng was supported by NSFC under Grant 81871508. The work of X. Liu was supported by NSFC under Grant 61876101. The work of D. Li was supported by NSFC under Grant 61773246. This article was recommended by Associate Editor D. Goldgof. (Corresponding author: Dinggang Shen.) J. Xue is with the Business School, Shandong Key Laboratory of Medical Physics and Image Processing, Shandong Normal University, Jinan 250014, China, and also with the Department of Radiology and Biomedical Research Imaging Center, University of North Carolina at Chapel Hill, Chapel Hill, NC 27510 USA.
Publisher Copyright:
© 2013 IEEE.
PY - 2021/4
Y1 - 2021/4
N2 - Automatic pancreas segmentation is crucial to the diagnostic assessment of diabetes or pancreatic cancer. However, the relatively small size of the pancreas in the upper body, as well as large variations of its location and shape in retroperitoneum, make the segmentation task challenging. To alleviate these challenges, in this article, we propose a cascaded multitask 3-D fully convolution network (FCN) to automatically segment the pancreas. Our cascaded network is composed of two parts. The first part focuses on fast locating the region of the pancreas, and the second part uses a multitask FCN with dense connections to refine the segmentation map for fine voxel-wise segmentation. In particular, our multitask FCN with dense connections is implemented to simultaneously complete tasks of the voxel-wise segmentation and skeleton extraction from the pancreas. These two tasks are complementary, that is, the extracted skeleton provides rich information about the shape and size of the pancreas in retroperitoneum, which can boost the segmentation of pancreas. The multitask FCN is also designed to share the low- and mid-level features across the tasks. A feature consistency module is further introduced to enhance the connection and fusion of different levels of feature maps. Evaluations on two pancreas datasets demonstrate the robustness of our proposed method in correctly segmenting the pancreas in various settings. Our experimental results outperform both baseline and state-of-the-art methods. Moreover, the ablation study shows that our proposed parts/modules are critical for effective multitask learning.
AB - Automatic pancreas segmentation is crucial to the diagnostic assessment of diabetes or pancreatic cancer. However, the relatively small size of the pancreas in the upper body, as well as large variations of its location and shape in retroperitoneum, make the segmentation task challenging. To alleviate these challenges, in this article, we propose a cascaded multitask 3-D fully convolution network (FCN) to automatically segment the pancreas. Our cascaded network is composed of two parts. The first part focuses on fast locating the region of the pancreas, and the second part uses a multitask FCN with dense connections to refine the segmentation map for fine voxel-wise segmentation. In particular, our multitask FCN with dense connections is implemented to simultaneously complete tasks of the voxel-wise segmentation and skeleton extraction from the pancreas. These two tasks are complementary, that is, the extracted skeleton provides rich information about the shape and size of the pancreas in retroperitoneum, which can boost the segmentation of pancreas. The multitask FCN is also designed to share the low- and mid-level features across the tasks. A feature consistency module is further introduced to enhance the connection and fusion of different levels of feature maps. Evaluations on two pancreas datasets demonstrate the robustness of our proposed method in correctly segmenting the pancreas in various settings. Our experimental results outperform both baseline and state-of-the-art methods. Moreover, the ablation study shows that our proposed parts/modules are critical for effective multitask learning.
KW - Multitask FCN
KW - pancreas segmentation
KW - skeleton extraction
UR - http://www.scopus.com/inward/record.url?scp=85084231170&partnerID=8YFLogxK
U2 - 10.1109/TCYB.2019.2955178
DO - 10.1109/TCYB.2019.2955178
M3 - Article
C2 - 31869812
AN - SCOPUS:85084231170
VL - 51
SP - 2153
EP - 2165
JO - IEEE Transactions on Cybernetics
JF - IEEE Transactions on Cybernetics
SN - 2168-2267
IS - 4
M1 - 8936540
ER -