University of Leicester
Browse
unsupervised_yilong(2).pdf (4.35 MB)

Du-net based Unsupervised Contrastive Learning for Cancer Segmentation in Histology Images

Download (4.35 MB)
conference contribution
posted on 2022-06-28, 09:11 authored by Y Li, Y Wang, Huiyu Zhou, H Wang, G Jia, Q Zhang

In this paper, we introduce an unsupervised cancer segmentation framework for histology images. The framework involves an effective contrastive learning scheme for extracting distinctive visual representations for segmentation. The encoder is a Deep U-Net (DU-Net) structure which contains an extra fully convolution layer compared to the normal U-Net. A contrastive learning scheme is developed to solve the problem of lacking training sets with high-quality annotations on tumour boundaries. A specific set of data augmentation techniques are employed to improve the discriminability of the learned colour features from contrastive learning. Smoothing and noise elimination are conducted using convolutional Conditional Random Fields. The experiments demonstrate competitive performance in segmentation even better than some popular supervised networks.

Funding

National Key Research and Development Program under Grant No. 2019YFC0118404, the Basic Public Welfare Research Project of Zhejiang Province (LGF20H180001)

History

Author affiliation

School of Computing and Mathematical Sciences, University of Leicester

Source

ICIRA 2022, The 15th International Conference on Intelligent Robotics and Applications. Harbin, China. August 1-3, 2022

Version

  • AM (Accepted Manuscript)

Published in

International Journal of Intelligent Robotics and Applications

Issue

ICIRA 2022.

Publisher

Springer

isbn

978-3-031-13840-9

Acceptance date

2022-06-13

Copyright date

2022

Available date

2023-09-15

Book series

Lecture Notes in Computer Science, vol 13458

Temporal coverage: start date

2022-08-01

Temporal coverage: end date

2022-08-03

Language

en

Publisher version

Usage metrics

    University of Leicester Publications

    Categories

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC