Related Experiment Video
Updated: Aug 14, 2025
08:12
A Volumetric Method for Quantification of Cerebral Vasospasm in a Murine Model of Subarachnoid Hemorrhage
Published on: July 28, 2018
8.1K
Deep learning-based semantic vessel graph extraction for intracranial aneurysm rupture risk management
Annika Niemann1,2, Daniel Behme3, Naomi Larsen4
1Department of Simulation and Graphics, Otto-von-Guericke University, Magdeburg, Germany.
International Journal of Computer Assisted Radiology and Surgery
|January 10, 2023
Summary
This study introduces an automated pipeline for analyzing intracranial aneurysms, improving rupture risk assessment. The deep learning approach enhances morphological analysis and achieves 83.3% accuracy in rupture state classification.
Area of Science:
- Medical Imaging
- Computational Fluid Dynamics
- Neurosurgery
Background:
- Intracranial aneurysms pose complex treatment challenges.
- Current rupture risk assessment is often oversimplified and unreliable, particularly for multiple aneurysms.
- Advanced intracranial aneurysm analysis requires extensive, complex preprocessing.
Purpose of the Study:
- To develop an automated pipeline for intracranial aneurysm analysis.
- To facilitate the clinical translation of advanced aneurysm research.
- To improve the accuracy of aneurysm rupture risk assessment.
Main Methods:
- A deep learning-based pipeline for mesh segmentation, centerline, and outlet detection.
- Automatic generation of a semantic vessel graph for morphological analysis.
- Rupture state classification using 3D surface models and semantic graphs.
Main Results:
- Successful application of deep learning for aneurysm surface mesh segmentation.
- Extraction of comprehensive morphological parameters using semantic vessel graphs.
- Classification accuracy of 83.3% for aneurysm rupture state.
- Identified slightly higher average torsion and curvature in vessels near ruptured aneurysms.
Conclusions:
- The pipeline automates intracranial aneurysm analysis with minimal user intervention.
- Semantic graph representation aids morphological and hemodynamical parameter extraction.
- The approach shows significant potential for deep learning-based rupture state classification.
