Risk prediction for elderly cognitive impairment by radiomic and morphological quantification analysis based on a...

Area of Science:

• Neurology
• Radiology
• Medical Imaging

Background:

• Cognitive impairment associated with cerebrovascular disease (CI-CVD) is a significant concern in the elderly population.
• Cerebral magnetic resonance angiography (MRA) offers a non-invasive method to visualize cerebral vasculature.
• Early detection of CI-CVD is crucial for timely intervention and management.

Purpose of the Study:

• To develop and validate morphological and radiomic models for the early prediction of CI-CVD using cerebral MRA data.
• To identify key morphological and radiomic features associated with CI-CVD in an elderly cohort.
• To assess the predictive performance of combined clinical, morphological, and radiomic models.

Main Methods:

• Retrospective analysis of a large elderly MRA cohort, including patients with CI-CVD and controls.
• Automated quantitative analysis of cerebral artery morphology (stenosis, length, angle, deviation).
• Screening of clinical and morphological risk factors using logistic regression and extraction of radiomic features using LASSO regression.

Main Results:

• History of stroke identified as a significant clinical risk factor for CI-CVD.
• Specific morphological features (e.g., RMCA stenosis, LICA deviation, RICA/LICA twisted angles) were significant predictors.
• A combined clinical-morphological-radiomic model achieved high predictive performance (AUC 0.883 in training, 0.843 in external validation).

Conclusions:

• Radiomic features and morphological indicators from cerebral MRA are effective for early CI-CVD detection in the elderly.
• The developed multipredictor model offers optimal performance for early warning of CI-CVD.
• Cerebral artery stenosis and tortuosity are key risk factors, highlighting the value of MRA-based analysis.