SE-U-Net: Contextual Segmentation by Loosely Coupled Deep Networks for Medical Imaging Industry

Lin Yi Jiang, Cheng Ju Kuo, O. Tang-Hsuan, Min Hsiung Hung, Chao Chun Chen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

We proposed a context segmentation method for medical images via two deep networks, aiming at providing segmentation contexts and achieving better segmentation quality. The context in this work means the object labels for segmentation. The key idea of our proposed scheme is to develop mechanisms to elegantly transform object detection labels into the segmentation network structure, so that two deep networks can collaboratively operate with loosely-coupled manner. For achieving this, the scalable data transformation mechanisms between two deep networks need to be invented, including representation of contexts obtained from the first deep network and context importion to the second one. The experimental results reveal that the proposed scheme indeed performs superior segmentation quality.

Original languageEnglish
Title of host publicationIntelligent Information and Database Systems - 13th Asian Conference, ACIIDS 2021, Proceedings
EditorsNgoc Thanh Nguyen, Suphamit Chittayasothorn, Dusit Niyato, Bogdan Trawiński
PublisherSpringer Science and Business Media Deutschland GmbH
Pages678-691
Number of pages14
ISBN (Print)9783030732790
DOIs
Publication statusPublished - 2021
Event13th Asian Conference on Intelligent Information and Database Systems, ACIIDS 2021 - Phuket, Thailand
Duration: 2021 Apr 72021 Apr 10

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12672 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference13th Asian Conference on Intelligent Information and Database Systems, ACIIDS 2021
Country/TerritoryThailand
CityPhuket
Period21-04-0721-04-10

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'SE-U-Net: Contextual Segmentation by Loosely Coupled Deep Networks for Medical Imaging Industry'. Together they form a unique fingerprint.

Cite this