OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA

Anushka A. Kore; Frank G. te Nijenhuis; Matthijs van der Sluijs; Wim van Zwam; Charles Majoie; Geert Lycklama à Nijeholt; Danny Ruijters; Frans Vos; Sandra Cornelissen; Ruisheng Su; Theo van Walsum

doi:10.1007/978-3-032-07945-9_3

OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA

Anushka A. Kore
, Frank G. te Nijenhuis^*
, Matthijs van der Sluijs
, Wim van Zwam
, Charles Majoie
, Geert Lycklama à Nijeholt
, Danny Ruijters
, Frans Vos
, Sandra Cornelissen
, Ruisheng Su
, Theo van Walsum

^*Corresponding author for this work

Delft University of Technology
Erasmus University Rotterdam
Maastricht University
Emma Center for Personalized Medicine
Haaglanden Clinics
TU Eindhoven

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

7 Downloads (Pure)

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.

Original language	English
Title of host publication	Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings
Editors	Ruisheng Su, Ezequiel de la Rosa, Linda Vorberg, Leonhard Rist, Jiong Zhang, Adam Hilbert, Theo van Walsum
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	22-31
Number of pages	10
Volume	16098 LNCS
ISBN (Print)	9783032079442
DOIs	https://doi.org/10.1007/978-3-032-07945-9_3
Publication status	Published - 2026
Event	5th International Workshop on Imaging and Treatment Challenges, SWITCH 2025, Held in Conjunction with Medical Image Computing and Computer Assisted Intervention, MICCAI 2025 - Daejeon, South Korea Duration: 23 Sept 2025 → 23 Sept 2025

Publication series

Name	Lecture Notes in Computer Science
Volume	16098 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	5th International Workshop on Imaging and Treatment Challenges, SWITCH 2025, Held in Conjunction with Medical Image Computing and Computer Assisted Intervention, MICCAI 2025
Country/Territory	South Korea
City	Daejeon
Period	23/09/2025 → 23/09/2025

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Acute Ischemic Stroke
Artificial Intelligence
Large Vessel Occlusion
Object Detection

Access to Document

10.1007/978-3-032-07945-9_3

Occlunet.pdfFinal published version, 13.4 MBLicence: Taverne

Cite this

Kore, A. A., te Nijenhuis, F. G., van der Sluijs, M., van Zwam, W., Majoie, C., Nijeholt, G. L. À., Ruijters, D., Vos, F., Cornelissen, S., Su, R., & van Walsum, T. (2026). OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA. In R. Su, E. de la Rosa, L. Vorberg, L. Rist, J. Zhang, A. Hilbert, & T. van Walsum (Eds.), Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings (Vol. 16098 LNCS, pp. 22-31). (Lecture Notes in Computer Science; Vol. 16098 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-032-07945-9_3

Kore, Anushka A. ; te Nijenhuis, Frank G. ; van der Sluijs, Matthijs et al. / OccluNet : Spatio-Temporal Deep Learning for Occlusion Detection on DSA. Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings. editor / Ruisheng Su ; Ezequiel de la Rosa ; Linda Vorberg ; Leonhard Rist ; Jiong Zhang ; Adam Hilbert ; Theo van Walsum. Vol. 16098 LNCS Springer Science and Business Media Deutschland GmbH, 2026. pp. 22-31 (Lecture Notes in Computer Science).

@inproceedings{ffdc4da18ef34bbd80fe8e8e05e4cbd1,

title = "OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA",

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{\textquoteright}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.",

keywords = "Acute Ischemic Stroke, Artificial Intelligence, Large Vessel Occlusion, Object Detection",

author = "Kore, \{Anushka A.\} and \{te Nijenhuis\}, \{Frank G.\} and \{van der Sluijs\}, Matthijs and \{van Zwam\}, Wim and Charles Majoie and Nijeholt, \{Geert Lycklama {\`a}\} and Danny Ruijters and Frans Vos and Sandra Cornelissen and Ruisheng Su and \{van Walsum\}, Theo",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 5th International Workshop on Imaging and Treatment Challenges, SWITCH 2025, Held in Conjunction with Medical Image Computing and Computer Assisted Intervention, MICCAI 2025 ; Conference date: 23-09-2025 Through 23-09-2025",

year = "2026",

doi = "10.1007/978-3-032-07945-9\_3",

language = "English",

isbn = "9783032079442",

volume = "16098 LNCS",

series = "Lecture Notes in Computer Science",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "22--31",

editor = "Ruisheng Su and \{de la Rosa\}, Ezequiel and Linda Vorberg and Leonhard Rist and Jiong Zhang and Adam Hilbert and \{van Walsum\}, Theo",

booktitle = "Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings",

address = "Germany",

}

Kore, AA, te Nijenhuis, FG, van der Sluijs, M, van Zwam, W, Majoie, C, Nijeholt, GLÀ, Ruijters, D, Vos, F, Cornelissen, S, Su, R & van Walsum, T 2026, OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA. in R Su, E de la Rosa, L Vorberg, L Rist, J Zhang, A Hilbert & T van Walsum (eds), Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings. vol. 16098 LNCS, Lecture Notes in Computer Science, vol. 16098 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 22-31, 5th International Workshop on Imaging and Treatment Challenges, SWITCH 2025, Held in Conjunction with Medical Image Computing and Computer Assisted Intervention, MICCAI 2025, Daejeon, South Korea, 23/09/2025. https://doi.org/10.1007/978-3-032-07945-9_3

OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA. / Kore, Anushka A.; te Nijenhuis, Frank G.; van der Sluijs, Matthijs et al.
Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings. ed. / Ruisheng Su; Ezequiel de la Rosa; Linda Vorberg; Leonhard Rist; Jiong Zhang; Adam Hilbert; Theo van Walsum. Vol. 16098 LNCS Springer Science and Business Media Deutschland GmbH, 2026. p. 22-31 (Lecture Notes in Computer Science; Vol. 16098 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - OccluNet

T2 - 5th International Workshop on Imaging and Treatment Challenges, SWITCH 2025, Held in Conjunction with Medical Image Computing and Computer Assisted Intervention, MICCAI 2025

AU - Kore, Anushka A.

AU - te Nijenhuis, Frank G.

AU - van der Sluijs, Matthijs

AU - van Zwam, Wim

AU - Majoie, Charles

AU - Nijeholt, Geert Lycklama à

AU - Ruijters, Danny

AU - Vos, Frans

AU - Cornelissen, Sandra

AU - Su, Ruisheng

AU - van Walsum, Theo

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

PY - 2026

Y1 - 2026

N2 - 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.

AB - 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.

KW - Acute Ischemic Stroke

KW - Artificial Intelligence

KW - Large Vessel Occlusion

KW - Object Detection

UR - https://www.scopus.com/pages/publications/105020018481

U2 - 10.1007/978-3-032-07945-9_3

DO - 10.1007/978-3-032-07945-9_3

M3 - Conference contribution

SN - 9783032079442

VL - 16098 LNCS

T3 - Lecture Notes in Computer Science

SP - 22

EP - 31

BT - Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings

A2 - Su, Ruisheng

A2 - de la Rosa, Ezequiel

A2 - Vorberg, Linda

A2 - Rist, Leonhard

A2 - Zhang, Jiong

A2 - Hilbert, Adam

A2 - van Walsum, Theo

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 23 September 2025 through 23 September 2025

ER -

Kore AA, te Nijenhuis FG, van der Sluijs M, van Zwam W, Majoie C, Nijeholt GLÀ et al. OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA. In Su R, de la Rosa E, Vorberg L, Rist L, Zhang J, Hilbert A, van Walsum T, editors, Image Analysis in Stroke Diagnosis and Interventions - 5th International Workshop, SWITCH 2025, Held in Conjunction with MICCAI 2025, Proceedings. Vol. 16098 LNCS. Springer Science and Business Media Deutschland GmbH. 2026. p. 22-31. (Lecture Notes in Computer Science). doi: 10.1007/978-3-032-07945-9_3

OccluNet: Spatio-Temporal Deep Learning for Occlusion Detection on DSA

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this