Related Experiment Video
Updated: Nov 12, 2025
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
43.1K
Multidimensional predicting model of intracranial aneurysm stability with backpropagation neural network: a
Yi Yang1,2, Qingyuan Liu1,2, Pengjun Jiang1,2
1Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, No. 119 South 4th Ring West Road, Fengtai District, Beijing, 100070, China.
Summary
This study shows that using multidimensional data with a Backpropagation (BP) neural network can accurately assess the 3-year risk of intracranial aneurysm (IA) rupture or growth. Multidimensional models significantly improve prediction accuracy compared to single or double-dimensional approaches.
Area of Science:
- Neurosurgery
- Medical Imaging
- Artificial Intelligence in Medicine
Background:
- Intracranial aneurysms (IAs) pose a rupture risk influenced by multiple factors.
- Backpropagation (BP) neural networks offer potential for clinical decision support in IA management.
Purpose of the Study:
- To evaluate the feasibility of BP neural networks for assessing IA rupture/growth risk.
- To demonstrate the superiority of multidimensional models over simpler models for IA stability prediction.
Main Methods:
- A prospective study included 36 IA patients with 37 IAs, followed for up to 36 months.
- Multidimensional data (clinical, morphological, hemodynamic) were collected and analyzed using patient-specific models.
- Seven BP neural network models were constructed and compared using Area Under the Curve (AUC) for predictive performance.
Main Results:
- Morphological characteristics alone showed moderate 3-year IA stability prediction (AUC = 0.703).
- Combined clinical-morphological, clinical-hemodynamic, and morphological-hemodynamic models demonstrated improved prediction (AUCs ranging from 0.702 to 0.785).
- The model integrating all three dimensions achieved the highest predictive significance for 3-year IA rupture/growth risk (AUC = 0.811, P = 0.001).
Conclusions:
- BP neural networks show promise for assessing 3-year IA stability.
- Integrating multidimensional patient data significantly enhances the accuracy of predicting IA rupture or growth risk.
