Publication Database

Comprehensive dataset of coarse tumor annotations for the cancer genome atlas breast invasive carcinoma (2025)

Art

Vortrag

Autoren

Banerjee, Sweta

Bertram, Christof A.

Ammeling, Jonas

Weiss, Viktoria

Conrad, Thomas (WE 12)

Klopfleisch, Robert (WE 12)

Kaltenecker, Christopher

Breininger, Katharina

Aubreville, Marc

Kongress

Bildverarbeitung für die Medizin 2025

Regensburg, 09. – 11.03.2025

Quelle

Bildverarbeitung für die Medizin 2025 : Proceedings, German Conference on Medical Image Computing, Regensburg March 09-11, 2025 — herausgegeben von Christoph Palm, Katharina Breininger, Thomas Deserno, Heinz Handels, Andreas Maier, Klaus H. Maier-Hein, Thomas M. Tolxdorff (Hrsg.)

1st Auflage

Wiesbaden: Springer Fachmedien Wiesbaden, Imprint: Springer Vieweg, 2025. Informatik aktuell — S. 260–265

ISBN: 978-3-658-47422-5

Sprache

Englisch

Verweise

URL (Volltext): https://doi.org/10.1007/978-3-658-47422-5

DOI: 10.1007/978-3-658-47422-5_56

Kontakt

Institut für Tierpathologie

Robert-von-Ostertag-Str. 15
14163 Berlin
+49 30 838 62450
pathologie@vetmed.fu-berlin.de

Abstract / Zusammenfassung

Automated tumor segmentation of histologic images is crucial for the development of computer-assisted diagnostic workflows aiming at accurate prognostication. We present a dataset of coarse annotations of over 1,000 highresolution breast tumor images from The cancer genome atlas breast invasive carcinoma (TCGA-BRCA) repository, each annotated with binary segmentation masks that delineate coarse tumor areas. Additionally, a subset of 20 images includes fine-grained annotations, providing precise delineation of tumor boundaries beyond the broader outlines used in coarse annotations. Initial evaluations using U-Net and DeepLabv3 models show promising segmentation performance. On a held-out, coarsely annotated test set, U-Net achieved an average intersection over union (IoU) score of 0.795, while DeepLabv3 scored 0.783. On the finely annotated subset of this test set, U-Net reached an average IoU of 0.746, with DeepLabv3 slightly outperforming at 0.765. The public availability of this dataset aims to support research in automated tumor analysis, advancing diagnostic workflows and thereby ultimately improving breast cancer management.