CSM-Net: A Multi-Task Colorectal Cancer Analysis Framework

Colorectal cancer is the second leading cause of cancer mortality before 2020, in this paper, we propose a simple yet efficient anchor-free instance segmentation framework, Colorectal Spatial Mask Network, aiming at jointly detecting and segmenting the tumor region in colorectal cancer MRI image ser...

Full description

Saved in:
Bibliographic Details
Published in:Sensing and imaging 2020-12, Vol.21 (1), Article 42
Main Authors: Sui, Dong, Zhang, Yimeng, Li, Zhaoxin, Zhao, Guangzhe, Wang, Yantao, Zhang, Rui, Wu, Qiufeng, Zhang, Lei
Format: Article
Language:English
Subjects:
Cancer
Colorectal cancer
Computer networks
Electrical Engineering
Engineering
Image segmentation
Imaging
Magnetic resonance imaging
Microwaves
Original Paper
Radiology
RF and Optical Engineering
S.I.: Deep Learning for Medical Image Reconstruction and Analysis
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Colorectal cancer is the second leading cause of cancer mortality before 2020, in this paper, we propose a simple yet efficient anchor-free instance segmentation framework, Colorectal Spatial Mask Network, aiming at jointly detecting and segmenting the tumor region in colorectal cancer MRI image series. we add a novel spatial attention-guided mask branch to anchor-free one stage object detector into the framework with the same way in Mask R-CNN. In addition, we also employ an improved network VoVNetV2 as the new backbone network. After comparison with the state-of-the-arts methods, the results of experiments reveal that our framework can get well balance between speed and accuracy.
ISSN:1557-2064
1557-2072
DOI:10.1007/s11220-020-00307-1