Related Experiment Video
Updated: May 19, 2026
03:47
A Murine Model of Carotid Aneurysm Formation
Published on: September 9, 2025
Multicenter-derived machine learning model for individualized rupture-risk prediction in intracranial aneurysms:
Manman Cui1, Shaokun Hu1, Yiru Shen1,2
1Department of Radiology, The Second Affiliated Hospital of Soochow University, Suzhou, China.
Quantitative Imaging in Medicine and Surgery
|May 18, 2026
Summary
Predicting intracranial aneurysm rupture risk is crucial for patient outcomes. This study developed an integrated model combining clinical, CTA morphological, and radiomics features, showing high accuracy and consistent performance across machine learning algorithms.
Area of Science:
- Neurosurgery
- Radiology
- Medical Imaging
- Machine Learning
Background:
- Intracranial aneurysms (IAs) cause most spontaneous subarachnoid hemorrhages (SAH), leading to significant mortality and morbidity.
- Predicting IA rupture risk is challenging due to inter-patient variability in influencing factors.
- Accurate risk stratification is essential for timely intervention and improved patient outcomes.
Purpose of the Study:
- To develop and evaluate a predictive model for intracranial aneurysm rupture risk.
- To integrate clinical, computed tomographic angiography (CTA) morphological, and radiomics features for enhanced prediction.
- To assess the performance consistency of various machine learning (ML) algorithms in predicting IA rupture.
Main Methods:
- A retrospective multicenter study enrolled 756 patients with 877 IAs.
- Clinical, morphological, and radiomics features were extracted and analyzed.
- Six predictive models were constructed using logistic regression, including a nomogram. Seven additional ML algorithms were applied for comparative analysis.
Main Results:
- Age, blood glucose, aneurysm diameter, location, parent artery diameter, and ten radiomics features were identified as independent predictors.
- The integrated clinical-morphological-radiomics model (Model_C&M&R) achieved high AUC values (0.887 training, 0.910 internal validation).
- The model demonstrated consistent generalizability in external validation sets (AUC 0.773 and 0.735).
Conclusions:
- The integrated Model_C&M&R offers optimal predictive accuracy for intracranial aneurysm rupture.
- Machine learning models developed using these features show consistently excellent performance.
- This integrated approach enhances the prediction of IA rupture risk, aiding clinical decision-making.
