Prediction of catheter navigation difficulty during mechanical thrombectomy using CTA-based carotid siphon
Ferenc Dezső Bakó1, Attila Tanács2, József Tolnai1
1Department of Medical Physics and Informatics, University of Szeged, Szeged, Hungary.
Frontiers in Medical Technology
|August 13, 2026
Summary
3D analysis of carotid siphon anatomy using CT angiography can predict device support needs during stroke thrombectomy. This approach improves accuracy in identifying cases requiring specific device escalation for successful catheter navigation.
Area of Science:
- Neurosurgery and Interventional Radiology
- Medical Imaging and Computational Anatomy
- Biomedical Engineering
Background:
- Mechanical thrombectomy for acute ischemic stroke faces challenges in navigating the carotid siphon.
- Complex vascular anatomy can necessitate device escalation, increasing procedure time and delaying reperfusion.
- Traditional 2D measurements may not fully represent 3D vascular features impacting device behavior.
Purpose of the Study:
- To evaluate if CT angiography (CTA)-derived carotid siphon morphometry can predict the device support level needed for aspiration catheter navigation.
- To assess the utility of 3D vascular analysis in optimizing procedural planning for mechanical thrombectomy.
Main Methods:
- Retrospective analysis of 53 patient CTA datasets undergoing endovascular thrombectomy.
- 3D segmentation and morphometric analysis of the carotid siphon using specialized software.
- Development of a logistic regression classifier to predict procedural difficulty based on geometric attributes.
Main Results:
- A model using CTA-derived morphometry achieved 75.0% accuracy in predicting device support needs in a high-confidence subset of 32 cases.
- The model demonstrated improved performance with a kappa statistic of 0.65 and a weighted F-measure of 0.73.
- High recall (100%) for M2 and precision (100%) for M3 categories were observed, though M4 remained challenging.
Conclusions:
- CTA-derived 3D carotid siphon morphometry offers clinically relevant insights into catheter navigation difficulty during thrombectomy.
- Confidence-gated classification enhances prediction reliability for anatomically defined cases.
- Further multicenter studies are needed to validate this predictive approach for patient-specific preprocedural planning.
