Representation disentanglement for multi-task learning with application to fetal ultrasound
Meng Q, Pawlowski N, Rueckert D, Kainz B (2019)
Publication Type: Conference contribution
Publication year: 2019
Journal
Publisher: Springer
Book Volume: 11798 LNCS
Pages Range: 47-55
Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Event location: Shenzhen, CHN
ISBN: 9783030328740
DOI: 10.1007/978-3-030-32875-7_6
Abstract
One of the biggest challenges for deep learning algorithms in medical image analysis is the indiscriminate mixing of image properties, e.g. artifacts and anatomy. These entangled image properties lead to a semantically redundant feature encoding for the relevant task and thus lead to poor generalization of deep learning algorithms. In this paper we propose a novel representation disentanglement method to extract semantically meaningful and generalizable features for different tasks within a multi-task learning framework. Deep neural networks are utilized to ensure that the encoded features are maximally informative with respect to relevant tasks, while an adversarial regularization encourages these features to be disentangled and minimally informative about irrelevant tasks. We aim to use the disentangled representations to generalize the applicability of deep neural networks. We demonstrate the advantages of the proposed method on synthetic data as well as fetal ultrasound images. Our experiments illustrate that our method is capable of learning disentangled internal representations. It outperforms baseline methods in multiple tasks, especially on images with new properties, e.g. previously unseen artifacts in fetal ultrasound.
Authors with CRIS profile
Involved external institutions
United Kingdom (GB)How to cite
APA:
Meng, Q., Pawlowski, N., Rueckert, D., & Kainz, B. (2019). Representation disentanglement for multi-task learning with application to fetal ultrasound. In Qian Wang, Alberto Gomez, Jana Hutter, Alberto Gomez, Veronika Zimmer, Jana Hutter, Emma Robinson, Daan Christiaens, Andrew Melbourne, Kristin McLeod, Oliver Zettinig, Roxane Licandro, Esra Abaci Turk (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 47-55). Shenzhen, CHN: Springer.
MLA:
Meng, Qingjie, et al. "Representation disentanglement for multi-task learning with application to fetal ultrasound." Proceedings of the 1st International Workshop on Smart Ultrasound Imaging, SUSI 2019, and the 4th International Workshop on Preterm, Perinatal and Paediatric Image Analysis, PIPPI 2019, held in conjunction with the 22nd International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2019, Shenzhen, CHN Ed. Qian Wang, Alberto Gomez, Jana Hutter, Alberto Gomez, Veronika Zimmer, Jana Hutter, Emma Robinson, Daan Christiaens, Andrew Melbourne, Kristin McLeod, Oliver Zettinig, Roxane Licandro, Esra Abaci Turk, Springer, 2019. 47-55.
BibTeX: Download