Fully Connected Crf With Data-driven Prior for Multi-class Brain Tumor Segmentation

Haocheng Shen; Jianguo Zhang

doi:10.1109/ICIP.2017.8296577

Fully Connected Crf With Data-driven Prior for Multi-class Brain Tumor Segmentation

Haocheng Shen, Jianguo Zhang

Computing

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

10 Citations (Scopus)

379 Downloads (Pure)

Abstract

Grid conditional random fields (CRFs) are widely applied in both natural and medical image segmentation tasks. However, they only consider the label coherence in neighborhood pixels or regions, which limits their ability to model long-range connections within the image and generally results in excessive smoothing of tumor boundaries. In this paper, we present a novel method for brain tumor segmentation in MR images based on fully-connected CRF (FC-CRF) model that establishes pairwise potentials on all pairs of pixels in the images. We employ a hierarchical approach to differentiate different structures of tumor and further formulate a FC-CRF model with learned data-driven prior knowledge of tumor core. The methods were evaluated on the testing and leaderboard set of Brain Tumor Image Segmentation Benchmark (BRATS) 2013 challenge. The precision of segmented tumor boundaries is improved significantly and the results are competitive compared to the start-of-the-arts.

Original language	English
Title of host publication	2017 IEEE International Conference on Image Processing (ICIP)
Subtitle of host publication	17-20 Sept. 2017
Publisher	IEEE
Pages	1727-1731
Number of pages	5
ISBN (Electronic)	9781509021758
ISBN (Print)	9781509021765
DOIs	https://doi.org/10.1109/ICIP.2017.8296577
Publication status	Published - 22 Feb 2018
Event	The International Conference on Image Processing 2017 - China National Convention Center, Beijing, China Duration: 17 Sept 2017 → 20 Dec 2017

Publication series

Name	IEEE Conference Proceedings
Publisher	IEEE
ISSN (Electronic)	2381-8549

Conference

Conference	The International Conference on Image Processing 2017
Abbreviated title	IEEE ICIP 2017
Country/Territory	China
City	Beijing
Period	17/09/17 → 20/12/17

Keywords

Tumors
Image segmentation
Brain modeling
Training
Testing
Three-dimensional displays
Kernel
CRF
Prior
Brain tumor segmentation

ASJC Scopus subject areas

Software
Signal Processing
Computer Vision and Pattern Recognition

Access to Document

10.1109/ICIP.2017.8296577

Author Accepted Manuscript
© 2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript, 405 KB

Cite this

@inproceedings{66a2942b2eac4462b19e27d1c7a719ce,

title = "Fully Connected Crf With Data-driven Prior for Multi-class Brain Tumor Segmentation",

abstract = "Grid conditional random fields (CRFs) are widely applied in both natural and medical image segmentation tasks. However, they only consider the label coherence in neighborhood pixels or regions, which limits their ability to model long-range connections within the image and generally results in excessive smoothing of tumor boundaries. In this paper, we present a novel method for brain tumor segmentation in MR images based on fully-connected CRF (FC-CRF) model that establishes pairwise potentials on all pairs of pixels in the images. We employ a hierarchical approach to differentiate different structures of tumor and further formulate a FC-CRF model with learned data-driven prior knowledge of tumor core. The methods were evaluated on the testing and leaderboard set of Brain Tumor Image Segmentation Benchmark (BRATS) 2013 challenge. The precision of segmented tumor boundaries is improved significantly and the results are competitive compared to the start-of-the-arts.",

keywords = "Tumors, Image segmentation, Brain modeling, Training, Testing, Three-dimensional displays, Kernel, CRF, Prior, Brain tumor segmentation",

author = "Haocheng Shen and Jianguo Zhang",

note = "This work was supported partially by the National Natural Science Foundation of China (No. 61628212).; The International Conference on Image Processing 2017, IEEE ICIP 2017 ; Conference date: 17-09-2017 Through 20-12-2017",

year = "2018",

month = feb,

day = "22",

doi = "10.1109/ICIP.2017.8296577",

language = "English",

isbn = "9781509021765",

series = "IEEE Conference Proceedings",

publisher = "IEEE",

pages = "1727--1731",

booktitle = "2017 IEEE International Conference on Image Processing (ICIP)",

}

TY - GEN

T1 - Fully Connected Crf With Data-driven Prior for Multi-class Brain Tumor Segmentation

AU - Shen, Haocheng

AU - Zhang, Jianguo

N1 - This work was supported partially by the National Natural Science Foundation of China (No. 61628212).

PY - 2018/2/22

Y1 - 2018/2/22

N2 - Grid conditional random fields (CRFs) are widely applied in both natural and medical image segmentation tasks. However, they only consider the label coherence in neighborhood pixels or regions, which limits their ability to model long-range connections within the image and generally results in excessive smoothing of tumor boundaries. In this paper, we present a novel method for brain tumor segmentation in MR images based on fully-connected CRF (FC-CRF) model that establishes pairwise potentials on all pairs of pixels in the images. We employ a hierarchical approach to differentiate different structures of tumor and further formulate a FC-CRF model with learned data-driven prior knowledge of tumor core. The methods were evaluated on the testing and leaderboard set of Brain Tumor Image Segmentation Benchmark (BRATS) 2013 challenge. The precision of segmented tumor boundaries is improved significantly and the results are competitive compared to the start-of-the-arts.

AB - Grid conditional random fields (CRFs) are widely applied in both natural and medical image segmentation tasks. However, they only consider the label coherence in neighborhood pixels or regions, which limits their ability to model long-range connections within the image and generally results in excessive smoothing of tumor boundaries. In this paper, we present a novel method for brain tumor segmentation in MR images based on fully-connected CRF (FC-CRF) model that establishes pairwise potentials on all pairs of pixels in the images. We employ a hierarchical approach to differentiate different structures of tumor and further formulate a FC-CRF model with learned data-driven prior knowledge of tumor core. The methods were evaluated on the testing and leaderboard set of Brain Tumor Image Segmentation Benchmark (BRATS) 2013 challenge. The precision of segmented tumor boundaries is improved significantly and the results are competitive compared to the start-of-the-arts.

KW - Tumors

KW - Image segmentation

KW - Brain modeling

KW - Training

KW - Testing

KW - Three-dimensional displays

KW - Kernel

KW - CRF

KW - Prior

KW - Brain tumor segmentation

UR - http://www.scopus.com/inward/record.url?scp=85045301362&partnerID=8YFLogxK

U2 - 10.1109/ICIP.2017.8296577

DO - 10.1109/ICIP.2017.8296577

M3 - Conference contribution

SN - 9781509021765

T3 - IEEE Conference Proceedings

SP - 1727

EP - 1731

BT - 2017 IEEE International Conference on Image Processing (ICIP)

PB - IEEE

T2 - The International Conference on Image Processing 2017

Y2 - 17 September 2017 through 20 December 2017

ER -

Fully Connected Crf With Data-driven Prior for Multi-class Brain Tumor Segmentation

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this