A Novel Radiomic Model for Risk Stratification of Cerebral Herniation in Radiation-Induced Cystic Brain Necrosis

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Radiation-induced cystic brain necrosis (RCN) poses a risk of rapid progression to life-threatening cerebral herniation.
• Early and accurate risk assessment is crucial for managing RCN patients.

Purpose of the Study:

• To develop and validate a predictive model for cerebral herniation in RCN patients.
• The model integrates radiomic features from MRI and clinical variables.

Main Methods:

• Retrospective analysis of 130 RCN patients treated for nasopharyngeal carcinoma.
• Extraction of radiomic features from baseline T2-weighted MRI.
• Development of a radiomic signature and a multivariate Cox regression model incorporating clinical variables.

Main Results:

• The radiomic signature demonstrated good predictive performance.
• The integrated radiomic model achieved high predictive accuracy (C-index: 0.841 training, 0.867 testing).
• The model effectively stratified patients into high- and low-risk groups with good calibration and clinical utility.

Conclusions:

• An MRI-based radiomic model integrating imaging and clinical data reliably predicts cerebral herniation in RCN patients.
• This model offers a practical tool to support clinical decision-making and improve patient outcomes.