Learning wall segmentation in 3D Vessel Trees using sparse annotations

Rahlfs, Hinrich; Hüllebrand, Markus; Schmitter, Sebastian; Strecker, Christoph; Harloff, Andreas; Hennemuth, Anja

doi:https://doi.org/10.1007/978-981-96-3863-5_52

Zugriffsnummer	54885
Dokumenttyp	Konferenzartikel
Sprache	Englisch
Titel	Learning wall segmentation in 3D Vessel Trees using sparse annotations
Autor(in); Institution	Rahlfs, Hinrich; Charite Univ Med Berlin, Inst Comp Assisted Cardiovasc Med, Berlin, GERMANY Hüllebrand, Markus; Charite Univ Med Berlin, Inst Comp Assisted Cardiovasc Med, Berlin, GERMANY; Fraunhofer MEVIS, Bremen, GERMANY Schmitter, Sebastian; 8.1, Biomedizinische Magnetresonanz, PTB-Berlin Strecker, Christoph; Univ Freiburg, Dept Neurol & Neurophysiol, Fac Med, Med Ctr, Freiburg Im Breisgau, GERMANY Harloff, Andreas; Univ Freiburg, Dept Neurol & Neurophysiol, Fac Med, Med Ctr, Freiburg Im Breisgau, GERMANY Hennemuth, Anja; Charite Univ Med Berlin, Inst Comp Assisted Cardiovasc Med, Berlin, GERMANY; Fraunhofer MEVIS, Bremen, GERMANY
Quelle/Jahr	Proceedings of 2024 International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2024):(2025), 569 - 578
ISSN	1876-1100 (print) ; 1876-1119 (online)
ISBN	978-981-96-3863-5
DOI	https://doi.org/10.1007/978-981-96-3863-5_52
URL	https://link.springer.com/content/pdf/10.1007/978-981-96-3863-5_52
Verlag	Singapore: Springer Nature Singapore
Konferenzangaben	International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2024), Manchester, 19-21, November, 2024, UK
Zusammenfassung	We propose a novel approach that uses sparse annotations from clinical studies to train a 3D segmentation of the carotid artery wall. We use a centerline annotation to sample perpendicular cross-sections of the carotid artery and use an adversarial 2D network to segment them. These annotations are then transformed into 3D pseudo-labels for training of a 3D convolutional neural network, circumventing the creation of manual 3D masks. For pseudo-label creation in the bifurcation area we propose the use of cross-sections perpendicular to the bifurcation axis and show that this enhances segmentation performance. Different sampling distances had a lesser impact. The proposed method allows for efficient training of 3D segmentation, offering potential improvements in the assessment of carotid artery stenosis and allowing the extraction of 3D biomarkers such as plaque volume.

Zitierung

Rahlfs, H., Hüllebrand, M., Schmitter, S., Strecker, C., Harloff, A., & Hennemuth, A. (2025). Learning wall segmentation in 3D Vessel Trees using sparse annotations. International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2024), Manchester, 19-21, November, 2024, UK. https://doi.org/10.1007/978-981-96-3863-5_52

Exportieren

RIS BibTeX

Zitierung

Exportieren

PTB-Publica Menü