Collaborative Multi-agent Reinforcement Learning for Landmark Localization Using Continuous Action Space

Klemens Kasseroller, Franz Thaler, Christian Payer, Darko Štern*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

Abstract

We propose a reinforcement learning (RL) based approach for anatomical landmark localization in medical images, where the agent can move in arbitrary directions with a variable step size. Using a continuous action space reduces the average number of steps required to locate a landmark by more than 30 times compared to localization using discrete actions. Our approach outperforms a state-of-the-art RL method based on a discrete action space and is inline with state-of-the-art supervised regression based methods. Furthermore, we extend our approach to a multi-agent setting, where we allow collaboration between agents to enable learning of the landmarks’ spatial configuration. The results of the multi-agent RL based approach show that the position of occluded landmarks can be successfully estimated based on the relative position predicted for the visible landmarks.

Original languageEnglish
Title of host publicationInformation Processing in Medical Imaging - 27th International Conference, IPMI 2021, Proceedings
EditorsAasa Feragen, Stefan Sommer, Julia Schnabel, Mads Nielsen
PublisherSpringer Science and Business Media Deutschland GmbH
Pages767-778
Number of pages12
ISBN (Print)9783030781903
DOIs
Publication statusPublished - 2021
Event27th International Conference on Information Processing in Medical Imaging, IPMI 2021 - Virtual, Online
Duration: 28 Jun 202130 Jun 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12729 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference27th International Conference on Information Processing in Medical Imaging, IPMI 2021
CityVirtual, Online
Period28/06/2130/06/21

Keywords

  • Collaborative multi-agent
  • Continuous action space
  • Landmark localization
  • Reinforcement learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Collaborative Multi-agent Reinforcement Learning for Landmark Localization Using Continuous Action Space'. Together they form a unique fingerprint.

Cite this