Vertebral artery fusiform aneurysm geometry in predicting rupture risk

Area of Science:

• Neurosurgery
• Medical Imaging
• Machine Learning

Background:

• Cerebral aneurysms are a widespread health concern.
• Predicting aneurysm rupture risk is crucial for clinical decision-making.
• Vertebral artery fusiform aneurysms (VAFAs) present unique challenges in risk assessment.

Purpose of the Study:

• To investigate if VAFA morphology predicts rupture risk.
• To identify key geometric features indicative of VAFA rupture.
• To develop a machine learning model for VAFA rupture risk classification.

Main Methods:

• Utilized image analysis and machine learning on 37 VAFA images (12 ruptured, 25 un-ruptured).
• Computed 571 geometric features and selected five statistically significant predictors.
• Applied a machine learning classification algorithm to assess VAFA morphology.

Main Results:

• Achieved state-of-the-art classification performance of 81.43 ± 13.08% for VAFA rupture risk.
• Identified five key geometric features predicting rupture risk: cross-sectional area change, distal vessel diameter, aneurysm solidity, distal vessel curvature, and proximal vessel curvature ratio.
• Demonstrated the potential of VAFA morphology as a non-invasive indicator.

Conclusions:

• Geometric features of VAFA morphology are significant predictors of rupture risk.
• Machine learning analysis of these features can aid in clinical management decisions.
• These findings suggest potential for non-invasive risk stratification in surgical settings.