TY - GEN
T1 - Multiple landmark detection using multi-agent reinforcement learning
AU - Vlontzos, Athanasios
AU - Alansary, Amir
AU - Kamnitsas, Konstantinos
AU - Rueckert, Daniel
AU - Kainz, Bernhard
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2019.
PY - 2019
Y1 - 2019
N2 - The detection of anatomical landmarks is a vital step for medical image analysis and applications for diagnosis, interpretation and guidance. Manual annotation of landmarks is a tedious process that requires domain-specific expertise and introduces inter-observer variability. This paper proposes a new detection approach for multiple landmarks based on multi-agent reinforcement learning. Our hypothesis is that the position of all anatomical landmarks is interdependent and non-random within the human anatomy, thus finding one landmark can help to deduce the location of others. Using a Deep Q-Network (DQN) architecture we construct an environment and agent with implicit inter-communication such that we can accommodate K agents acting and learning simultaneously, while they attempt to detect K different landmarks. During training the agents collaborate by sharing their accumulated knowledge for a collective gain. We compare our approach with state-of-the-art architectures and achieve significantly better accuracy by reducing the detection error by 50%, while requiring fewer computational resources and time to train compared to the naïve approach of training K agents separately. Code and visualizations available: https://github.com/thanosvlo/MARL-for-Anatomical-Landmark-Detection.
AB - The detection of anatomical landmarks is a vital step for medical image analysis and applications for diagnosis, interpretation and guidance. Manual annotation of landmarks is a tedious process that requires domain-specific expertise and introduces inter-observer variability. This paper proposes a new detection approach for multiple landmarks based on multi-agent reinforcement learning. Our hypothesis is that the position of all anatomical landmarks is interdependent and non-random within the human anatomy, thus finding one landmark can help to deduce the location of others. Using a Deep Q-Network (DQN) architecture we construct an environment and agent with implicit inter-communication such that we can accommodate K agents acting and learning simultaneously, while they attempt to detect K different landmarks. During training the agents collaborate by sharing their accumulated knowledge for a collective gain. We compare our approach with state-of-the-art architectures and achieve significantly better accuracy by reducing the detection error by 50%, while requiring fewer computational resources and time to train compared to the naïve approach of training K agents separately. Code and visualizations available: https://github.com/thanosvlo/MARL-for-Anatomical-Landmark-Detection.
UR - http://www.scopus.com/inward/record.url?scp=85075670029&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-32251-9_29
DO - 10.1007/978-3-030-32251-9_29
M3 - Conference contribution
AN - SCOPUS:85075670029
SN - 9783030322502
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 262
EP - 270
BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2019 - 22nd International Conference, Proceedings
A2 - Shen, Dinggang
A2 - Yap, Pew-Thian
A2 - Liu, Tianming
A2 - Peters, Terry M.
A2 - Khan, Ali
A2 - Staib, Lawrence H.
A2 - Essert, Caroline
A2 - Zhou, Sean
PB - Springer Science and Business Media Deutschland GmbH
T2 - 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019
Y2 - 13 October 2019 through 17 October 2019
ER -