Identification of Small, Regularly Shaped Cerebral Aneurysms Prone to Rupture

Area of Science:

• Neurosurgery
• Biomedical Engineering
• Radiology

Background:

• Many small, regularly shaped cerebral aneurysms rupture despite low scores from current risk-assessment methods.
• Existing risk-assessment tools may underestimate the rupture potential of small, regularly shaped aneurysms.
• There is a need to identify specific characteristics that predict rupture in this subpopulation.

Purpose of the Study:

• To identify patient and aneurysm characteristics associated with rupture in small, regularly shaped cerebral aneurysms.
• To develop and validate predictive models for aneurysm rupture in this specific group.
• To improve risk stratification for small, regularly shaped cerebral aneurysms.

Main Methods:

• Machine learning models were trained using data from 1079 small (<7 mm), regularly shaped cerebral aneurysms.
• Predictive models incorporated patient, aneurysm location, hemodynamic, and geometric features derived from computational fluid dynamics.
• Model performance was validated on an independent dataset of 102 small, regularly shaped aneurysms.

Main Results:

• Adverse hemodynamic environments (e.g., high speed, complex flow, oscillatory wall shear stress) were linked to rupture.
• Ruptured aneurysms were larger and more elongated than unruptured ones in this subset.
• The best machine learning model achieved an area under the curve of 0.84 for rupture prediction on validation data.

Conclusions:

• Predictive machine learning models can effectively identify small, regularly shaped aneurysms at high risk of rupture.
• Integrating hemodynamic, geometric, and anatomic data enhances rupture risk prediction.
• These models offer a promising approach to personalized risk assessment for cerebral aneurysms.