Development and validation of a radiomics model based on the ASPECTS framework using CT imaging for predicting...

Area of Science:

• Neurology
• Radiology
• Artificial Intelligence

Background:

• Accurate infarct delineation in acute stroke CT is challenging.
• Radiomics applications for stroke prediction are limited.
• Manual infarct segmentation is time-consuming and requires expertise.

Purpose of the Study:

• To develop and validate a multimodal prediction model for malignant cerebral edema (MCE).
• To utilize radiomic features extracted using the Alberta Stroke Program Early Computed Tomography Score (ASPECTS) framework.
• To eliminate the need for manual infarct segmentation in MCE prediction.

Main Methods:

• Retrospective multicenter study of acute ischemic stroke (AIS) patients.
• Extraction of radiomic features from ASPECTS regions on NCCT and CTA.
• Development and validation of fused clinical, imaging, and radiomic models using machine learning.

Main Results:

• The fused model achieved superior predictive performance across training, internal, and external validation cohorts (AUCs ranging from 0.78 to 0.91).
• Key predictors included ASPECTS, NIHSS score, and collateral score.
• The fused model demonstrated high specificity (82.5%) and accuracy (78.3%).

Conclusions:

• A fused model integrating clinical, radiological, and radiomic features shows superior and generalizable MCE prediction.
• The ASPECTS-based framework avoids manual segmentation, enabling rapid risk estimation.
• This model can be integrated into clinical workflows for timely MCE risk assessment.