Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning

Xiaojun Ma; Haixia Liu; Yanxiong Niu; Chengfen Zhang; Di Liu

doi:10.1109/ISCID.2018.10163

Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning

Xiaojun Ma
, Haixia Liu^*
, Yanxiong Niu
, Chengfen Zhang
, Di Liu

^*此作品的通讯作者

仪器科学与光电工程学院

Beihang University

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

3 引用（Scopus）

摘要

The recognition and classification of whole-slide pathological images is the core technology of computer-aided diagnosis of cancer. This paper proposes a new data set construction method to improve the computer-aided diagnosis method based on deep learning. Taking the pathological image of breast cancer as an example, the image was preprocessed by the otsu algorithm, and the amount of calculation was reduced by removing the blank area caused by tissue slide production. The noise picture database was established, and the data set was established by two different division strategies. The pre-trained googlenet model is used to train it; then the trained model is used to classify and diagnose pathological images. The experimental results show that the model AUC of the improved data set reaches 0.8410. An improved whole-slide pathology image recognition method based on deep learning is expected to be more widely used in clinical practice, making cancer diagnosis more efficient.

源语言	英语
主期刊名	Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018
出版商	Institute of Electrical and Electronics Engineers Inc.
页	269-272
页数	4
ISBN（电子版）	9781538685266
DOI	https://doi.org/10.1109/ISCID.2018.10163
出版状态	已出版 - 2 7月 2018
活动	11th International Symposium on Computational Intelligence and Design, ISCID 2018 - Hangzhou, 中国期限: 8 12月 2018 → 9 12月 2018

出版系列

姓名	Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018
卷	2

会议

会议	11th International Symposium on Computational Intelligence and Design, ISCID 2018
国家/地区	中国
市	Hangzhou
时期	8/12/18 → 9/12/18

联合国可持续发展目标

此成果有助于实现下列可持续发展目标：

可持续发展目标 3 良好健康与福祉

访问文件

10.1109/ISCID.2018.10163

其它文件与链接

链接到 Scopus 的出版物

引用此

Ma, X., Liu, H., Niu, Y., Zhang, C., & Liu, D. (2018). Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning. 在 Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018 (页码 269-272). 文章 8695469 (Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018; 卷 2). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISCID.2018.10163

Ma, Xiaojun ; Liu, Haixia ; Niu, Yanxiong 等. / Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning. Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018. Institute of Electrical and Electronics Engineers Inc., 2018. 页码 269-272 (Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018).

@inproceedings{ee6cbd4376444335b4157bdc5f3ad583,

title = "Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning",

abstract = "The recognition and classification of whole-slide pathological images is the core technology of computer-aided diagnosis of cancer. This paper proposes a new data set construction method to improve the computer-aided diagnosis method based on deep learning. Taking the pathological image of breast cancer as an example, the image was preprocessed by the otsu algorithm, and the amount of calculation was reduced by removing the blank area caused by tissue slide production. The noise picture database was established, and the data set was established by two different division strategies. The pre-trained googlenet model is used to train it; then the trained model is used to classify and diagnose pathological images. The experimental results show that the model AUC of the improved data set reaches 0.8410. An improved whole-slide pathology image recognition method based on deep learning is expected to be more widely used in clinical practice, making cancer diagnosis more efficient.",

keywords = "Cancer, Component, Deep learning, Pathological image, Transfer learning, Whole-slide image",

author = "Xiaojun Ma and Haixia Liu and Yanxiong Niu and Chengfen Zhang and Di Liu",

note = "Publisher Copyright: {\textcopyright} 2018 IEEE.; 11th International Symposium on Computational Intelligence and Design, ISCID 2018 ; Conference date: 08-12-2018 Through 09-12-2018",

year = "2018",

month = jul,

day = "2",

doi = "10.1109/ISCID.2018.10163",

language = "英语",

series = "Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "269--272",

booktitle = "Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018",

address = "美国",

}

Ma, X, Liu, H , Niu, Y, Zhang, C & Liu, D 2018, Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning. 在 Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018., 8695469, Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018, 卷 2, Institute of Electrical and Electronics Engineers Inc., 页码 269-272, 11th International Symposium on Computational Intelligence and Design, ISCID 2018, Hangzhou, 中国, 8/12/18. https://doi.org/10.1109/ISCID.2018.10163

Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning. / Ma, Xiaojun; Liu, Haixia ; Niu, Yanxiong 等.
Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018. Institute of Electrical and Electronics Engineers Inc., 2018. 页码 269-272 8695469 (Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018; 卷 2).

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

TY - GEN

T1 - Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning

AU - Ma, Xiaojun

AU - Liu, Haixia

AU - Niu, Yanxiong

AU - Zhang, Chengfen

AU - Liu, Di

PY - 2018/7/2

Y1 - 2018/7/2

N2 - The recognition and classification of whole-slide pathological images is the core technology of computer-aided diagnosis of cancer. This paper proposes a new data set construction method to improve the computer-aided diagnosis method based on deep learning. Taking the pathological image of breast cancer as an example, the image was preprocessed by the otsu algorithm, and the amount of calculation was reduced by removing the blank area caused by tissue slide production. The noise picture database was established, and the data set was established by two different division strategies. The pre-trained googlenet model is used to train it; then the trained model is used to classify and diagnose pathological images. The experimental results show that the model AUC of the improved data set reaches 0.8410. An improved whole-slide pathology image recognition method based on deep learning is expected to be more widely used in clinical practice, making cancer diagnosis more efficient.

AB - The recognition and classification of whole-slide pathological images is the core technology of computer-aided diagnosis of cancer. This paper proposes a new data set construction method to improve the computer-aided diagnosis method based on deep learning. Taking the pathological image of breast cancer as an example, the image was preprocessed by the otsu algorithm, and the amount of calculation was reduced by removing the blank area caused by tissue slide production. The noise picture database was established, and the data set was established by two different division strategies. The pre-trained googlenet model is used to train it; then the trained model is used to classify and diagnose pathological images. The experimental results show that the model AUC of the improved data set reaches 0.8410. An improved whole-slide pathology image recognition method based on deep learning is expected to be more widely used in clinical practice, making cancer diagnosis more efficient.

KW - Cancer

KW - Component

KW - Deep learning

KW - Pathological image

KW - Transfer learning

KW - Whole-slide image

UR - https://www.scopus.com/pages/publications/85065394849

U2 - 10.1109/ISCID.2018.10163

DO - 10.1109/ISCID.2018.10163

M3 - 会议稿件

AN - SCOPUS:85065394849

T3 - Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018

SP - 269

EP - 272

BT - Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 11th International Symposium on Computational Intelligence and Design, ISCID 2018

Y2 - 8 December 2018 through 9 December 2018

ER -

Ma X, Liu H , Niu Y, Zhang C, Liu D. Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning. 在 Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018. Institute of Electrical and Electronics Engineers Inc. 2018. 页码 269-272. 8695469. (Proceedings - 2018 11th International Symposium on Computational Intelligence and Design, ISCID 2018). doi: 10.1109/ISCID.2018.10163

Improvement of Whole-Slide Pathological Image Recognition Method Based on Deep Learning

摘要

出版系列

会议

联合国可持续发展目标

访问文件

其它文件与链接

学术指纹

引用此