Publication
Interpretable Mediastinal Lymph Node Station Classification and N-staging on CT and PET/CT Images
Sofija Engelson; Jan Ehrhardt; Yannic Elser; Malte M. Sieren; Julia Andresen; Stefanie Schierholz; Tobias Keck; Daniel Drömann; Jörg Barkhausen; Heinz Handels
In: Heinz Handels; Katharina Breininger; Thomas Deserno; Andreas Maier; Klaus Maier-Hein; Christoph Palm; Thomas Tolxdorff (Hrsg.). Bildverarbeitung für die Medizin 2026. German Conference on Medical Image Computing (BVM-2026), located at BVM-2026, March 15-17, Lübeck, Germany, Pages 1-9, ISBN 978-3-658-51100-5, Springer Fachmedien Wiesbaden, 3/2026.
Abstract
We present an interpretable approach for automated lymph node station (LNS) classification and N-staging on PET/CT and CT only by extending two established segmentation algorithms with probabilistic atlas-based LNS mapping. Our results show that a probabilistic approach for LNS mapping improves the detection accuracy by over 40 percentage points. The proposed method yields an accuracy of 0.74 for LNS classification and 0.68 for N-staging on PET/CT, representing a significant improvement toward human-level performance compared with the baseline approach. A performance drop for CT only evaluation indicates the PET scan adds valuable information to lymph node assessment, which is in alignment with according literature.
