Classification of skin pigmented lesions based on deep residual network

Yunfei Qi; Shaofu Lin; Zhisheng Huang

doi:10.1007/978-3-030-32962-4_6

Classification of skin pigmented lesions based on deep residual network

Yunfei Qi^*
, Shaofu Lin
, Zhisheng Huang

^*Corresponding author for this work

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

52 Downloads (Pure)

Abstract

There are various of skin pigmented lesions with high risk. Melanoma is one of the most dangerous forms of skin cancer. It is one of the important research directions of medical artificial intelligence to carry out classification research of skin pigmented lesions based on deep learning. It can assist doctors to make clinical diagnosis and make patients receive treatment as soon as possible to improve survival rate. Aiming at the similar and imbalanced dermoscopic image data of pigmented lesions, this paper proposes a deep residual network improved by Squeeze-and-Excitation module, and dynamic update class-weight, in batches, with model ensemble adjustment strategies to change the attention of imbalanced data. The results show that the above method can increase the average precision by 9.1%, the average recall by 15.3%, and the average F1-score by 12.2%, compared with the multi-class classification using the deep residual network. Thus, the above method is a better classification model and weight adjustment strategy.

Original language	English
Title of host publication	Health Information Science
Subtitle of host publication	8th International Conference, HIS 2019, Xi'an, China, October 18–20, 2019, Proceedings
Editors	Hua Wang, Siuly Siuly, Yanchun Zhang, Rui Zhou, Fernando Martin-Sanchez, Zhisheng Huang
Publisher	Springer
Pages	58-67
Number of pages	10
ISBN (Electronic)	9783030329624
ISBN (Print)	9783030329617
DOIs	https://doi.org/10.1007/978-3-030-32962-4_6
Publication status	Published - 2019
Event	8th International Conference on Health Information Science, HIS 2019 - Xi'an, China Duration: 18 Oct 2019 → 20 Oct 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11837 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	8th International Conference on Health Information Science, HIS 2019
Country/Territory	China
City	Xi'an
Period	18/10/19 → 20/10/19

Funding

Acknowledgments. This study was financially supported by program Research on Artificial Intelligence Innovation Technology for Mental Health Service, which is funded by the Beijing High-level Foreign Talents Subsidy Program 2019. The program number is Z201919. Our team continues to conduct research on artificial intelligence and big data analytics in the medical field, hoping to help human health with the power of data. And we are grateful to all study participants. This study was financially supported by program Research on Artificial Intelligence Innovation Technology for Mental Health Service, which is funded by the Beijing High-level Foreign Talents Subsidy Program 2019. The program number is Z201919. Our team continues to conduct research on artificial intelligence and big data analytics in the medical field, hoping to help human health with the power of data. And we are grateful to all study participants.

Funders	Funder number
program Research on Artificial Intelligence Innovation Technology for Mental Health Service	Z201919

UN SDGs

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

SDG 3 Good Health and Well-being

Keywords

Deep learning
Imbalanced data
Model ensemble
Multi-classification
Residual network
Skin lesions

Access to Document

10.1007/978-3-030-32962-4_6

Classification of Skin Pigmented Lesions Based on Deep Residual NetworkFinal published version, 354 KBLicence: Article 25fa Dutch Copyright Act

Persistent URL (handle)

Persistent link

Cite this

Qi, Y., Lin, S., & Huang, Z. (2019). Classification of skin pigmented lesions based on deep residual network. In H. Wang, S. Siuly, Y. Zhang, R. Zhou, F. Martin-Sanchez, & Z. Huang (Eds.), Health Information Science: 8th International Conference, HIS 2019, Xi'an, China, October 18–20, 2019, Proceedings (pp. 58-67). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11837 LNCS). Springer. https://doi.org/10.1007/978-3-030-32962-4_6

Qi, Yunfei ; Lin, Shaofu ; Huang, Zhisheng. / Classification of skin pigmented lesions based on deep residual network. Health Information Science: 8th International Conference, HIS 2019, Xi'an, China, October 18–20, 2019, Proceedings. editor / Hua Wang ; Siuly Siuly ; Yanchun Zhang ; Rui Zhou ; Fernando Martin-Sanchez ; Zhisheng Huang. Springer, 2019. pp. 58-67 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{5b44f9a46cb44d88a8ce960b670c4b05,

title = "Classification of skin pigmented lesions based on deep residual network",

abstract = "There are various of skin pigmented lesions with high risk. Melanoma is one of the most dangerous forms of skin cancer. It is one of the important research directions of medical artificial intelligence to carry out classification research of skin pigmented lesions based on deep learning. It can assist doctors to make clinical diagnosis and make patients receive treatment as soon as possible to improve survival rate. Aiming at the similar and imbalanced dermoscopic image data of pigmented lesions, this paper proposes a deep residual network improved by Squeeze-and-Excitation module, and dynamic update class-weight, in batches, with model ensemble adjustment strategies to change the attention of imbalanced data. The results show that the above method can increase the average precision by 9.1\%, the average recall by 15.3\%, and the average F1-score by 12.2\%, compared with the multi-class classification using the deep residual network. Thus, the above method is a better classification model and weight adjustment strategy.",

keywords = "Deep learning, Imbalanced data, Model ensemble, Multi-classification, Residual network, Skin lesions",

author = "Yunfei Qi and Shaofu Lin and Zhisheng Huang",

year = "2019",

doi = "10.1007/978-3-030-32962-4\_6",

language = "English",

isbn = "9783030329617",

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

publisher = "Springer",

pages = "58--67",

editor = "Hua Wang and Siuly Siuly and Yanchun Zhang and Rui Zhou and Fernando Martin-Sanchez and Zhisheng Huang",

booktitle = "Health Information Science",

note = "8th International Conference on Health Information Science, HIS 2019 ; Conference date: 18-10-2019 Through 20-10-2019",

}

Qi, Y, Lin, S & Huang, Z 2019, Classification of skin pigmented lesions based on deep residual network. in H Wang, S Siuly, Y Zhang, R Zhou, F Martin-Sanchez & Z Huang (eds), Health Information Science: 8th International Conference, HIS 2019, Xi'an, China, October 18–20, 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11837 LNCS, Springer, pp. 58-67, 8th International Conference on Health Information Science, HIS 2019, Xi'an, China, 18/10/19. https://doi.org/10.1007/978-3-030-32962-4_6

Classification of skin pigmented lesions based on deep residual network. / Qi, Yunfei; Lin, Shaofu; Huang, Zhisheng.
Health Information Science: 8th International Conference, HIS 2019, Xi'an, China, October 18–20, 2019, Proceedings. ed. / Hua Wang; Siuly Siuly; Yanchun Zhang; Rui Zhou; Fernando Martin-Sanchez; Zhisheng Huang. Springer, 2019. p. 58-67 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11837 LNCS).

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

TY - GEN

T1 - Classification of skin pigmented lesions based on deep residual network

AU - Qi, Yunfei

AU - Lin, Shaofu

AU - Huang, Zhisheng

PY - 2019

Y1 - 2019

N2 - There are various of skin pigmented lesions with high risk. Melanoma is one of the most dangerous forms of skin cancer. It is one of the important research directions of medical artificial intelligence to carry out classification research of skin pigmented lesions based on deep learning. It can assist doctors to make clinical diagnosis and make patients receive treatment as soon as possible to improve survival rate. Aiming at the similar and imbalanced dermoscopic image data of pigmented lesions, this paper proposes a deep residual network improved by Squeeze-and-Excitation module, and dynamic update class-weight, in batches, with model ensemble adjustment strategies to change the attention of imbalanced data. The results show that the above method can increase the average precision by 9.1%, the average recall by 15.3%, and the average F1-score by 12.2%, compared with the multi-class classification using the deep residual network. Thus, the above method is a better classification model and weight adjustment strategy.

AB - There are various of skin pigmented lesions with high risk. Melanoma is one of the most dangerous forms of skin cancer. It is one of the important research directions of medical artificial intelligence to carry out classification research of skin pigmented lesions based on deep learning. It can assist doctors to make clinical diagnosis and make patients receive treatment as soon as possible to improve survival rate. Aiming at the similar and imbalanced dermoscopic image data of pigmented lesions, this paper proposes a deep residual network improved by Squeeze-and-Excitation module, and dynamic update class-weight, in batches, with model ensemble adjustment strategies to change the attention of imbalanced data. The results show that the above method can increase the average precision by 9.1%, the average recall by 15.3%, and the average F1-score by 12.2%, compared with the multi-class classification using the deep residual network. Thus, the above method is a better classification model and weight adjustment strategy.

KW - Deep learning

KW - Imbalanced data

KW - Model ensemble

KW - Multi-classification

KW - Residual network

KW - Skin lesions

UR - https://www.scopus.com/pages/publications/85075648754

UR - https://www.scopus.com/pages/publications/85075648754#tab=citedBy

U2 - 10.1007/978-3-030-32962-4_6

DO - 10.1007/978-3-030-32962-4_6

M3 - Conference contribution

AN - SCOPUS:85075648754

SN - 9783030329617

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

SP - 58

EP - 67

BT - Health Information Science

A2 - Wang, Hua

A2 - Siuly, Siuly

A2 - Zhang, Yanchun

A2 - Zhou, Rui

A2 - Martin-Sanchez, Fernando

A2 - Huang, Zhisheng

PB - Springer

T2 - 8th International Conference on Health Information Science, HIS 2019

Y2 - 18 October 2019 through 20 October 2019

ER -

Qi Y, Lin S, Huang Z. Classification of skin pigmented lesions based on deep residual network. In Wang H, Siuly S, Zhang Y, Zhou R, Martin-Sanchez F, Huang Z, editors, Health Information Science: 8th International Conference, HIS 2019, Xi'an, China, October 18–20, 2019, Proceedings. Springer. 2019. p. 58-67. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). Epub 2019 Oct 8. doi: 10.1007/978-3-030-32962-4_6

Classification of skin pigmented lesions based on deep residual network

Abstract

Publication series

Conference

Funding

UN SDGs

Keywords

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this