OccluNet

Spatio-Temporal Deep Learning for Occlusion Detection on DSA

Conference Paper (2026)
Author(s)

Anushka A. Kore (Student TU Delft)

Frank G. te Nijenhuis (Erasmus MC)

Matthijs van der Sluijs (Erasmus MC)

Wim van Zwam (Maastricht University Medical Center)

Charles Majoie (Amsterdam UMC)

Geert Lycklama à. Nijeholt (Haaglanden Medical Center)

Danny Ruijters (Eindhoven University of Technology)

Frans Vos (TU Delft - ImPhys/Computational Imaging, TU Delft - ImPhys/Vos group)

Sandra Cornelissen (Erasmus MC)

Ruisheng Su (Eindhoven University of Technology)

Theo van Walsum (Erasmus MC)

DOI related publication
https://doi.org/10.1007/978-3-032-07945-9_3 Final published version
More Info
expand_more
Publication Year
2026
Language
English
Bibliographical Note
Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/publishing/publisher-deals Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.
Pages (from-to)
22-31
Publisher
Springer
ISBN (print)
9783032079442
Event
Downloads counter
144
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Accurate detection of vascular occlusions during endovascular thrombectomy (EVT) is critical in acute ischemic stroke (AIS). Interpretation of digital subtraction angiography (DSA) sequences poses challenges due to anatomical complexity and time constraints. This work proposes OccluNet, a spatio-temporal deep learning model that integrates YOLOX, a single-stage object detector, with transformer-based temporal attention mechanisms to automate occlusion detection in DSA sequences. We compared OccluNet with a YOLOv11 baseline trained on either individual DSA frames or minimum intensity projections. Two spatio-temporal variants were explored for OccluNet: pure temporal attention and divided space-time attention. Evaluation on DSA images from the MR CLEAN Registry revealed the model’s capability to capture temporally consistent features, achieving precision and recall of 89.02% and 74.87%, respectively. OccluNet significantly outperformed the baseline models, and both attention variants attained similar performance. Source code is available here.

Files

Taverne78-3-032-07945-9_3.pdf
(pdf | 2.02 Mb)
- Embargo expired in 10-04-2026
License info not available