Data Augmentation Based on DiscrimDiff for Histopathology Image Classification

Xianchao Guan; Yifeng Wang; Yiyang Lin; Yongbing Zhang

doi:10.1007/978-3-031-58171-7_6

Data Augmentation Based on DiscrimDiff for Histopathology Image Classification

Xianchao Guan
, Yifeng Wang
, Yiyang Lin
, Yongbing Zhang^*

^*Corresponding author for this work

Harbin Institute of Technology Shenzhen
Tsinghua University

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

2 Scopus citations

Abstract

Histopathological analysis is the present gold standard for cancer diagnosis. Accurate classification of histopathology images has great clinical significance and application value for assisting pathologists in diagnosis. However, the performance of histopathology image classification is greatly affected by data imbalance. To address this problem, we propose a novel data augmentation framework based on the diffusion model, DiscrimDiff, which expands the dataset by synthesizing images of rare classes. To compensate for the lack of discrimination ability of the diffusion model for synthesized images, we design a post-discrimination mechanism to provide image quality assurance for data augmentation. Our method significantly improves classification performance on multiple datasets. Furthermore, histomorphological features of different classes concerned by the diffusion model may provide guiding significance for pathologists in clinical diagnosis. Therefore, we visualize histomorphological features related to classification, which can be used to assist pathologist-in-training education and improve the understanding of histomorphology.

Original language	English
Title of host publication	Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings
Editors	Yuan Xue, Chen Chen, Chao Chen, Lianrui Zuo, Yihao Liu
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	53-62
Number of pages	10
ISBN (Print)	9783031581700
DOIs	https://doi.org/10.1007/978-3-031-58171-7_6
State	Published - 2024
Externally published	Yes
Event	3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023 - Vancouver, Canada Duration: 12 Oct 2023 → 12 Oct 2023

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14379 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023
Country/Territory	Canada
City	Vancouver
Period	12/10/23 → 12/10/23

UN SDGs

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

SDG 3 Good Health and Well-being

Keywords

Computational pathology
Data augmentation
Diffusion models
Histomorphological features

Access to Document

10.1007/978-3-031-58171-7_6

Cite this

Guan, X., Wang, Y., Lin, Y., & Zhang, Y. (2024). Data Augmentation Based on DiscrimDiff for Histopathology Image Classification. In Y. Xue, C. Chen, C. Chen, L. Zuo, & Y. Liu (Eds.), Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings (pp. 53-62). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14379 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-58171-7_6

Guan, Xianchao ; Wang, Yifeng ; Lin, Yiyang et al. / Data Augmentation Based on DiscrimDiff for Histopathology Image Classification. Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. editor / Yuan Xue ; Chen Chen ; Chao Chen ; Lianrui Zuo ; Yihao Liu. Springer Science and Business Media Deutschland GmbH, 2024. pp. 53-62 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{886b77f847cd49fc814ffb75c579c539,

title = "Data Augmentation Based on DiscrimDiff for Histopathology Image Classification",

abstract = "Histopathological analysis is the present gold standard for cancer diagnosis. Accurate classification of histopathology images has great clinical significance and application value for assisting pathologists in diagnosis. However, the performance of histopathology image classification is greatly affected by data imbalance. To address this problem, we propose a novel data augmentation framework based on the diffusion model, DiscrimDiff, which expands the dataset by synthesizing images of rare classes. To compensate for the lack of discrimination ability of the diffusion model for synthesized images, we design a post-discrimination mechanism to provide image quality assurance for data augmentation. Our method significantly improves classification performance on multiple datasets. Furthermore, histomorphological features of different classes concerned by the diffusion model may provide guiding significance for pathologists in clinical diagnosis. Therefore, we visualize histomorphological features related to classification, which can be used to assist pathologist-in-training education and improve the understanding of histomorphology.",

keywords = "Computational pathology, Data augmentation, Diffusion models, Histomorphological features",

author = "Xianchao Guan and Yifeng Wang and Yiyang Lin and Yongbing Zhang",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023 ; Conference date: 12-10-2023 Through 12-10-2023",

year = "2024",

doi = "10.1007/978-3-031-58171-7\_6",

language = "英语",

isbn = "9783031581700",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "53--62",

editor = "Yuan Xue and Chen Chen and Chao Chen and Lianrui Zuo and Yihao Liu",

booktitle = "Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings",

address = "德国",

}

Guan, X, Wang, Y, Lin, Y & Zhang, Y 2024, Data Augmentation Based on DiscrimDiff for Histopathology Image Classification. in Y Xue, C Chen, C Chen, L Zuo & Y Liu (eds), Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14379 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 53-62, 3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023, Vancouver, Canada, 12/10/23. https://doi.org/10.1007/978-3-031-58171-7_6

Data Augmentation Based on DiscrimDiff for Histopathology Image Classification. / Guan, Xianchao; Wang, Yifeng; Lin, Yiyang et al.
Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. ed. / Yuan Xue; Chen Chen; Chao Chen; Lianrui Zuo; Yihao Liu. Springer Science and Business Media Deutschland GmbH, 2024. p. 53-62 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14379 LNCS).

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

TY - GEN

T1 - Data Augmentation Based on DiscrimDiff for Histopathology Image Classification

AU - Guan, Xianchao

AU - Wang, Yifeng

AU - Lin, Yiyang

AU - Zhang, Yongbing

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

PY - 2024

Y1 - 2024

N2 - Histopathological analysis is the present gold standard for cancer diagnosis. Accurate classification of histopathology images has great clinical significance and application value for assisting pathologists in diagnosis. However, the performance of histopathology image classification is greatly affected by data imbalance. To address this problem, we propose a novel data augmentation framework based on the diffusion model, DiscrimDiff, which expands the dataset by synthesizing images of rare classes. To compensate for the lack of discrimination ability of the diffusion model for synthesized images, we design a post-discrimination mechanism to provide image quality assurance for data augmentation. Our method significantly improves classification performance on multiple datasets. Furthermore, histomorphological features of different classes concerned by the diffusion model may provide guiding significance for pathologists in clinical diagnosis. Therefore, we visualize histomorphological features related to classification, which can be used to assist pathologist-in-training education and improve the understanding of histomorphology.

AB - Histopathological analysis is the present gold standard for cancer diagnosis. Accurate classification of histopathology images has great clinical significance and application value for assisting pathologists in diagnosis. However, the performance of histopathology image classification is greatly affected by data imbalance. To address this problem, we propose a novel data augmentation framework based on the diffusion model, DiscrimDiff, which expands the dataset by synthesizing images of rare classes. To compensate for the lack of discrimination ability of the diffusion model for synthesized images, we design a post-discrimination mechanism to provide image quality assurance for data augmentation. Our method significantly improves classification performance on multiple datasets. Furthermore, histomorphological features of different classes concerned by the diffusion model may provide guiding significance for pathologists in clinical diagnosis. Therefore, we visualize histomorphological features related to classification, which can be used to assist pathologist-in-training education and improve the understanding of histomorphology.

KW - Computational pathology

KW - Data augmentation

KW - Diffusion models

KW - Histomorphological features

UR - https://www.scopus.com/pages/publications/85192908767

U2 - 10.1007/978-3-031-58171-7_6

DO - 10.1007/978-3-031-58171-7_6

M3 - 会议稿件

AN - SCOPUS:85192908767

SN - 9783031581700

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

SP - 53

EP - 62

BT - Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings

A2 - Xue, Yuan

A2 - Chen, Chen

A2 - Chen, Chao

A2 - Zuo, Lianrui

A2 - Liu, Yihao

PB - Springer Science and Business Media Deutschland GmbH

T2 - 3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023

Y2 - 12 October 2023 through 12 October 2023

ER -

Guan X, Wang Y, Lin Y, Zhang Y. Data Augmentation Based on DiscrimDiff for Histopathology Image Classification. In Xue Y, Chen C, Chen C, Zuo L, Liu Y, editors, Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. Springer Science and Business Media Deutschland GmbH. 2024. p. 53-62. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-58171-7_6