Domain Adversarial RetinaNet as a Reference Algorithm for the MItosis DOmain Generalization Challenge

Wilm F, Marzahl C, Breininger K, Aubreville M (2022)

Publication Type: Conference contribution

Publication year: 2022

Journal

Lecture Notes in Computer Science Springer Verlag

Publisher: Springer Science and Business Media Deutschland GmbH

Book Volume: 13166 LNCS

Pages Range: 5-13

Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Event location: Strasbourg, FRA

ISBN: 9783030972806

DOI: 10.1007/978-3-030-97281-3_1

Abstract

Assessing the mitotic count has a known high degree of intra- and inter-rater variability. Computer-aided systems have proven to decrease this variability and reduce labeling time. These systems, however, are generally highly dependent on their training domain and show poor applicability to unseen domains. In histopathology, these domain shifts can result from various sources, including different slide scanning systems used to digitize histologic samples. The MItosis DOmain Generalization challenge focused on this specific domain shift for the task of mitotic figure detection. This work presents a mitotic figure detection algorithm developed as a baseline for the challenge, based on domain adversarial training. On the challenge’s test set, the algorithm scored an F 1 score of 0.7183. The corresponding network weights and code for implementing the network are made publicly available.

Authors with CRIS profile

Frauke Wilm Lehrstuhl für Informatik 5 (Mustererkennung) Christian Marzahl Lehrstuhl für Informatik 5 (Mustererkennung) Katharina Breininger Lehrstuhl für Informatik 14 (Bild- und Sprachverarbeitung)

Involved external institutions

Technische Hochschule Ingolstadt DE Germany (DE)

How to cite

APA:

Wilm, F., Marzahl, C., Breininger, K., & Aubreville, M. (2022). Domain Adversarial RetinaNet as a Reference Algorithm for the MItosis DOmain Generalization Challenge. In Marc Aubreville, David Zimmerer, Mattias Heinrich (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 5-13). Strasbourg, FRA: Springer Science and Business Media Deutschland GmbH.

MLA:

Wilm, Frauke, et al. "Domain Adversarial RetinaNet as a Reference Algorithm for the MItosis DOmain Generalization Challenge." Proceedings of the 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021, Strasbourg, FRA Ed. Marc Aubreville, David Zimmerer, Mattias Heinrich, Springer Science and Business Media Deutschland GmbH, 2022. 5-13.

BibTeX: Download