Pathological Image Contrastive Self-supervised Learning

Qin，Wenkang1; Jiang，Shan2; Luo，Lin1,3

10.1007/978-3-031-16876-5_9

2022

1st International Workshop on Resource-Efficient Medical Image Analysis (REMIA) / 25th International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI)

0302-9743

1611-3349

978-3-031-16875-8

LECT NOTES ARTIF INT

13543 LNCS

85-94

SEP 22, 2022

null,Singapore,SINGAPORE

GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND

SPRINGER INTERNATIONAL PUBLISHING AG

Self-supervised learning methods have been receiving wide attentions in recent years, where contrastive learning starts to show encouraging performance in many tasks in the field of computer vision. Contrastive learning methods build pre-training weight parameters by crafting positive/negative samples and optimizing their distance in the feature space. It is easy to construct positive/negative samples on natural images, but the methods cannot directly apply to histopathological images because of the unique characteristics of the images such as staining invariance and vertical flip invariance. This paper proposes a general method for constructing clinical-equivalent positive sample pairs on histopathological images for applying contrastive learning on histopathological images. Results on the PatchCamelyon benchmark show that our method can improve model accuracy up to 6% while reducing the training costs, as well as reducing reliance on labeled data.

Histopathological images Representation learning Self-supervised learning

Corresponding

English

CPCI-S

Foundation of Shenzhen Science and Technology Innovation Committee[JCYJ20180507181527806]

Computer Science ; Pathology ; Radiology, Nuclear Medicine & Medical Imaging

Computer Science, Artificial Intelligence ; Pathology ; Radiology, Nuclear Medicine & Medical Imaging

WOS:000869764400009

2-s2.0-85138783901

Scopus

1.College of Engineering,Peking University,Beijing,China
2.Beijing Institute of Collaborative Innovation,Beijing,China
3.Southern University of Science and Technology,Shenzhen,China

Qin，Wenkang,Jiang，Shan,Luo，Lin. Pathological Image Contrastive Self-supervised Learning[C]. GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND:SPRINGER INTERNATIONAL PUBLISHING AG,2022:85-94.