MRI Radiomic Signature of White Matter Hyperintensities Is Associated With Clinical Phenotypes.

Area of Science:

• Neuroimaging
• Radiology
• Stroke Research

Background:

• Brain structural integrity is crucial for health, often assessed with advanced neuroimaging.
• Radiomics offers a quantitative method to analyze conventional images for brain health evaluation.
• White matter hyperintensities (WMH) are key indicators of brain damage.

Purpose of the Study:

• To evaluate radiomics for assessing brain structural integrity by predicting WMH burden.
• To explore associations between radiomic features and clinical phenotypes in acute ischemic stroke (AIS) patients.

Main Methods:

• Analysis of T2-FLAIR MR images from 4,163 AIS patients across multiple sites.
• Extraction of radiomic features from normal-appearing brain tissue.
• Prediction of WMH burden using ElasticNet linear regression and development of a radiomic signature.
• Canonical correlation analysis (CCA) to link the radiomic signature with clinical variables.

Main Results:

• Radiomic features effectively predicted WMH burden (R² = 0.855 ± 0.011).
• Significant correlations were found between the WMH radiomic signature and clinical traits (e.g., age, sex, smoking, diabetes, hypertension, AF, CAD).
• Seven canonical variates revealed distinct associations between radiomic patterns and clinical factors.

Conclusions:

• Radiomics from T2-FLAIR images capture cerebral microstructural damage in AIS patients.
• Radiomic features correlate with clinical phenotypes, indicating textural abnormalities related to cardiovascular risk.
• Radiomics shows promise for predicting WMH progression and monitoring stroke patients' brain health.