Pulmonary DR Image Anomaly Detection Based on Deep Learning

Zhendong Song; Lei Fan; Dong Huang; Xiaoyi Feng

doi:10.1007/978-3-030-34120-6_15

Pulmonary DR Image Anomaly Detection Based on Deep Learning

Zhendong Song
, Lei Fan
, Dong Huang
, Xiaoyi Feng

School of Electronics and Information

Northwestern Polytechnical University Xian

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

Abstract

The morbidity and mortality in lung cancer is increasing which makes the diagnosis of abnormal lungs particularly important. Because of the advantages in DR image, this paper aimed at two problems in current medical image research: first, it is difficult to completely segment the lung of DR image only used traditional image segmentation methods. This paper replaces the padding in the U-net network model with zero padding to maintain the image size and apply it to the lung DR image segmentation, and finally uses the lung DR image dataset to fine-tuning. Secondly, the results of anomaly detection experiments show that the algorithm would get more complete segmentation of lung DR images. Secondly, because of the insufficient of training set, the idea of multi-classifier fusion is used. Combining Gabor-based SVM classification, 3D convolutional neural network, and transfer learning to achieve a more complete description of features and make full use of the classification advantages of multi-classifiers. The experimental results show that the classification accuracy of this algorithm is 6% higher than that of the Transfer-ImageNet algorithm, 5% higher than SVM, 15% higher than 3D convolutional neural network, and improved 2.5% compared with FT-Transfer-DenseNet3D algorithm.

Original language	English
Title of host publication	Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1
Editors	Yao Zhao, Chunyu Lin, Nick Barnes, Baoquan Chen, Rüdiger Westermann, Xiangwei Kong
Publisher	Springer
Pages	182-198
Number of pages	17
ISBN (Print)	9783030341190
DOIs	https://doi.org/10.1007/978-3-030-34120-6_15
State	Published - 2019
Event	10th International Conference on Image and Graphics, ICIG 2019 - Beijing, China Duration: 23 Aug 2019 → 25 Aug 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11901 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	10th International Conference on Image and Graphics, ICIG 2019
Country/Territory	China
City	Beijing
Period	23/08/19 → 25/08/19

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

3D convolutional neural network
DR image
Lung field segmentation
SVM
Transfer learning

Access to Document

10.1007/978-3-030-34120-6_15

Cite this

Song, Z., Fan, L., Huang, D., & Feng, X. (2019). Pulmonary DR Image Anomaly Detection Based on Deep Learning. In Y. Zhao, C. Lin, N. Barnes, B. Chen, R. Westermann, & X. Kong (Eds.), Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1 (pp. 182-198). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11901 LNCS). Springer. https://doi.org/10.1007/978-3-030-34120-6_15

Song, Zhendong ; Fan, Lei ; Huang, Dong et al. / Pulmonary DR Image Anomaly Detection Based on Deep Learning. Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1. editor / Yao Zhao ; Chunyu Lin ; Nick Barnes ; Baoquan Chen ; Rüdiger Westermann ; Xiangwei Kong. Springer, 2019. pp. 182-198 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c7dad98f0ef54e37928801f52dcdb897,

title = "Pulmonary DR Image Anomaly Detection Based on Deep Learning",

abstract = "The morbidity and mortality in lung cancer is increasing which makes the diagnosis of abnormal lungs particularly important. Because of the advantages in DR image, this paper aimed at two problems in current medical image research: first, it is difficult to completely segment the lung of DR image only used traditional image segmentation methods. This paper replaces the padding in the U-net network model with zero padding to maintain the image size and apply it to the lung DR image segmentation, and finally uses the lung DR image dataset to fine-tuning. Secondly, the results of anomaly detection experiments show that the algorithm would get more complete segmentation of lung DR images. Secondly, because of the insufficient of training set, the idea of multi-classifier fusion is used. Combining Gabor-based SVM classification, 3D convolutional neural network, and transfer learning to achieve a more complete description of features and make full use of the classification advantages of multi-classifiers. The experimental results show that the classification accuracy of this algorithm is 6\% higher than that of the Transfer-ImageNet algorithm, 5\% higher than SVM, 15\% higher than 3D convolutional neural network, and improved 2.5\% compared with FT-Transfer-DenseNet3D algorithm.",

keywords = "3D convolutional neural network, DR image, Lung field segmentation, SVM, Transfer learning",

author = "Zhendong Song and Lei Fan and Dong Huang and Xiaoyi Feng",

note = "Publisher Copyright: {\textcopyright} 2019, Springer Nature Switzerland AG.; 10th International Conference on Image and Graphics, ICIG 2019 ; Conference date: 23-08-2019 Through 25-08-2019",

year = "2019",

doi = "10.1007/978-3-030-34120-6\_15",

language = "英语",

isbn = "9783030341190",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "182--198",

editor = "Yao Zhao and Chunyu Lin and Nick Barnes and Baoquan Chen and R{\"u}diger Westermann and Xiangwei Kong",

booktitle = "Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1",

}

Song, Z, Fan, L, Huang, D & Feng, X 2019, Pulmonary DR Image Anomaly Detection Based on Deep Learning. in Y Zhao, C Lin, N Barnes, B Chen, R Westermann & X Kong (eds), Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11901 LNCS, Springer, pp. 182-198, 10th International Conference on Image and Graphics, ICIG 2019, Beijing, China, 23/08/19. https://doi.org/10.1007/978-3-030-34120-6_15

Pulmonary DR Image Anomaly Detection Based on Deep Learning. / Song, Zhendong; Fan, Lei; Huang, Dong et al.
Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1. ed. / Yao Zhao; Chunyu Lin; Nick Barnes; Baoquan Chen; Rüdiger Westermann; Xiangwei Kong. Springer, 2019. p. 182-198 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11901 LNCS).

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

TY - GEN

T1 - Pulmonary DR Image Anomaly Detection Based on Deep Learning

AU - Song, Zhendong

AU - Fan, Lei

AU - Huang, Dong

AU - Feng, Xiaoyi

PY - 2019

Y1 - 2019

N2 - The morbidity and mortality in lung cancer is increasing which makes the diagnosis of abnormal lungs particularly important. Because of the advantages in DR image, this paper aimed at two problems in current medical image research: first, it is difficult to completely segment the lung of DR image only used traditional image segmentation methods. This paper replaces the padding in the U-net network model with zero padding to maintain the image size and apply it to the lung DR image segmentation, and finally uses the lung DR image dataset to fine-tuning. Secondly, the results of anomaly detection experiments show that the algorithm would get more complete segmentation of lung DR images. Secondly, because of the insufficient of training set, the idea of multi-classifier fusion is used. Combining Gabor-based SVM classification, 3D convolutional neural network, and transfer learning to achieve a more complete description of features and make full use of the classification advantages of multi-classifiers. The experimental results show that the classification accuracy of this algorithm is 6% higher than that of the Transfer-ImageNet algorithm, 5% higher than SVM, 15% higher than 3D convolutional neural network, and improved 2.5% compared with FT-Transfer-DenseNet3D algorithm.

AB - The morbidity and mortality in lung cancer is increasing which makes the diagnosis of abnormal lungs particularly important. Because of the advantages in DR image, this paper aimed at two problems in current medical image research: first, it is difficult to completely segment the lung of DR image only used traditional image segmentation methods. This paper replaces the padding in the U-net network model with zero padding to maintain the image size and apply it to the lung DR image segmentation, and finally uses the lung DR image dataset to fine-tuning. Secondly, the results of anomaly detection experiments show that the algorithm would get more complete segmentation of lung DR images. Secondly, because of the insufficient of training set, the idea of multi-classifier fusion is used. Combining Gabor-based SVM classification, 3D convolutional neural network, and transfer learning to achieve a more complete description of features and make full use of the classification advantages of multi-classifiers. The experimental results show that the classification accuracy of this algorithm is 6% higher than that of the Transfer-ImageNet algorithm, 5% higher than SVM, 15% higher than 3D convolutional neural network, and improved 2.5% compared with FT-Transfer-DenseNet3D algorithm.

KW - 3D convolutional neural network

KW - DR image

KW - Lung field segmentation

KW - SVM

KW - Transfer learning

UR - https://www.scopus.com/pages/publications/85076922407

U2 - 10.1007/978-3-030-34120-6_15

DO - 10.1007/978-3-030-34120-6_15

M3 - 会议稿件

AN - SCOPUS:85076922407

SN - 9783030341190

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 182

EP - 198

BT - Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1

A2 - Zhao, Yao

A2 - Lin, Chunyu

A2 - Barnes, Nick

A2 - Chen, Baoquan

A2 - Westermann, Rüdiger

A2 - Kong, Xiangwei

PB - Springer

T2 - 10th International Conference on Image and Graphics, ICIG 2019

Y2 - 23 August 2019 through 25 August 2019

ER -

Song Z, Fan L, Huang D, Feng X. Pulmonary DR Image Anomaly Detection Based on Deep Learning. In Zhao Y, Lin C, Barnes N, Chen B, Westermann R, Kong X, editors, Image and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1. Springer. 2019. p. 182-198. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-34120-6_15

Pulmonary DR Image Anomaly Detection Based on Deep Learning

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this