Predictive models for assessing the risk of brain aneurysm rupture

Area of Science:

• Neurosurgery
• Radiology
• Medical Imaging Analysis

Background:

• Intracranial aneurysm instability, indicated by increased wall enhancement, poses a significant clinical challenge.
• High-resolution magnetic resonance imaging (HR-MRI) offers detailed visualization of aneurysm morphology and wall characteristics.

Purpose of the Study:

• To evaluate and compare the performance of different predictive models for intracranial aneurysm instability.
• To determine the added value of aneurysm wall enhancement and radiomics in predicting symptomatic presentation.

Main Methods:

• Prospective HR-MRI scans were performed on patients with intracranial aneurysms.
• Aneurysm instability was defined as rupture or symptomatic status.
• Predictive models were developed using clinical data (PHASES score), morphological metrics, aneurysm wall enhancement, and radiomics features.

Main Results:

• The best performing model, incorporating age and radiomic data, achieved an AUC of 0.87, with 76% accuracy, 88% sensitivity, and 72% specificity.
• Models combining clinical data with morphological metrics and wall enhancement showed progressively improved predictive performance.
• The PHASES score alone had an AUC of 0.62, while adding morphological data improved it to 0.79.

Conclusions:

• Personalized triage of intracranial aneurysms can be significantly enhanced by integrating clinical data, detailed morphological analysis, and advanced imaging techniques like radiomics.
• Sophisticated analysis of aneurysm wall enhancement provides crucial insights into aneurysm instability.
• Radiomics features, combined with clinical data, represent a promising tool for predicting aneurysm presentation.