Shanghai Institute of Technology Zhang Xiaobing, Nie Shengdong, etc.: Three-dimensional segmentation of glioma integrating multiple densely connected two-dimensional convolutional neural network segmentation models

Content Introduction Chinese Abstract: Summary: In order to overcome the computational burden of processing three-dimensional medical scans and the lack of spatial information in two-dimensional medical scans, this paper proposes a new segmentation method for glioma, which combines three densely connected two The segmentation results of the two-dimensional convolutional neural network (2D-CNN) segmentation model are fused
.

In order to combine low-level features and high-level features, this paper adds tightly connected modules to the network structure design, so that as the number of network layers increases during the learning process, low-level features will not be missed
.

In addition, in order to solve the problem of blurring the boundary of the glioma edema area, we superimposed and fused the fluid attenuation inversion recovery (FLAIR) modal image and the T2 modal image to enhance the edema area
.

For the loss function of network training, we improved the cross-entropy loss function to effectively avoid network overfitting
.

In this paper, experiments are performed on the multi-modal brain tumor image segmentation dataset (BraTS)
.

Among them, our method reached 0.
84, 0.
82 and 0.
83 on the BraTS2018 training set for the entire tumor, tumor core and enhanced tumor area, respectively; on the BraTS2018 validation set, it reached 0.
82, 0.
85 and 0.
83; on the BraTS2013 test set Achieved 0.
81, 0.
78 and 0.
83
.

The experimental results show that the method has good accuracy and fast processing ability, and has a good clinical application prospect
.

Keyword group: glioma; magnetic resonance imaging (MRI); segmentation; densely connected blocks; two-dimensional convolutional neural networks (2D-CNNs) Author: Xiaobing ZHANG, Yin HU, Wen CHEN, Gang HUANG, Shengdong NIE Format: Xiaobing ZHANG, Yin HU, Wen CHEN, Gang HUANG, Shengdong NIE.
3D brain glioma segmentation in MRI through integrating multiple densely connected 2D convolutional neural networks[J].
Journal of Zhejiang University Science B, 2021, 22(6): 462-475.
https://doi.
org/10.
1631/jzus.
B2000381 This article is an essential guide: Full text download link: http:// jzus.
B2000381 About this journal "Journal of Zhejiang University (English Edition) Series B: Biomedicine and Biotechnology", referred to as JZUS-B, is a comprehensive English monthly magazine for biology
.

JZUS-B since 2008 were included SCI-E, the latest impact factor of 3.
066, while being MEDLINE / PubMed, PMC, Scopus, BA, BIOSIS Previews other well-known platforms retrieval
.

The scope of articles accepted by JZUS-B covers life sciences, medicine, pharmacy, biochemistry, biotechnology and bioengineering, agronomy, ecological environment, and zoology and other related disciplines
.

Article types include Review, Article, Perspective, Viewpoint, Editorial, Correspondence, etc.

.

The journal has always implemented a strict international peer review mechanism, with an average review period of 42 days and an average acceptance time of 53 days
.

Articles will be in press online immediately after acceptance, and will generally be officially published within 2 months
.

"Content-oriented, flexible form" is the characteristic of this journal
.

JZUS-B welcomes outstanding scholars at home and abroad to contribute the latest scientific research results, controversial topics, review papers, and organize the publication of hot albums
.

Journal Springer homepage: https:// Online submission address: http:// Contact Us The English version of Zhejiang University News is a number of English academic journals The collection scope covers all directions of science, engineering, agriculture and medicine.
Welcome to contribute to our journal.
You can follow our public account zdxbywb for more information
.