Advancing Intracranial Aneurysm Risk Stratification with Radiomics

Clinicians are increasingly seeking quantitative tools to enhance intracranial aneurysm risk stratification. While high-resolution vessel wall imaging (HR-VWI) provides valuable qualitative insights, researchers are now exploring its quantitative potential. Consequently, a recent study developed a radiomics-based model that significantly improves the identification of symptomatic intracranial aneurysms (SIAs).

Improving Intracranial Aneurysm Risk Stratification with Radiomics

Traditional risk assessments, like the PHASES score, often rely on morphological parameters. However, these methods sometimes lack the sensitivity needed for precise clinical management. Specifically, researchers in this study utilized HR-VWI to extract detailed radiomic features from 446 aneurysms. Notably, they examined both the aneurysm wall and the parent artery wall to capture a comprehensive vascular profile. Similarly, this approach allowed the team to identify specific imaging biomarkers associated with symptomatic status.

The study specifically developed multiple models, including RadscoreIA and RadscorePA. Furthermore, they created a combined model, RadscoreLOCIA_PA, which integrated aneurysm location as a critical variable. Indeed, this comprehensive tool demonstrated remarkable accuracy during validation. In fact, the RadscoreLOCIA_PA model achieved an area under the curve (AUC) of 0.888. This result clearly outperformed the traditional PHASES score, which only reached an AUC of 0.679. Moreover, the new model showed superior calibration and clinical benefit.

Additionally, the integration of parent artery features proved essential. Most previous studies focused solely on the aneurysm sac itself. However, by including the parent artery wall, the researchers captured broader hemodynamic and inflammatory influences. Consequently, this multi-region analysis provides a more robust framework for intracranial aneurysm risk stratification. Therefore, clinicians can rely on these models for more nuanced diagnostics and treatment planning.

Clinical Implications and Future Practice

Finally, the findings suggest that radiomics can transform how neurologists manage patients. This objective method reduces reliance on subjective qualitative assessment. Thus, clinicians can make better-informed decisions regarding intervention. As these models undergo further validation, they will likely become a standard component of neuroimaging protocols. Undoubtedly, such advancements improve patient outcomes through personalized care strategies.

FAQs

What is the PHASES score?

The PHASES score is a traditional clinical tool used to estimate the five-year risk of aneurysm rupture based on factors like age, hypertension, and aneurysm size.

How does radiomics improve risk assessment?

Radiomics extracts hundreds of quantitative data points from medical images that are invisible to the naked eye, allowing for more precise characterization of tissue pathology.

Why was the parent artery included in this model?

The parent artery wall often exhibits inflammatory changes that reflect the overall stability of the vascular environment, making it a valuable predictor for symptomatic status.

Disclaimer: This content is for informational and educational purposes only. It is not intended to be a substitute for professional medical advice, diagnosis, or treatment. Always seek the advice of your physician or other qualified health provider with any questions you may have regarding a medical condition. Refer to the latest local and national guidelines for clinical practice.

References

Zhou Z et al. Radiomics model for risk stratification of intracranial aneurysm: a high-resolution vessel wall imaging-based study. BMC Med Imaging. 2026 May 12. doi: 10.1186/s12880-026-02417-w. PMID: 42120992.

Greving JP et al. Predictors of de novo aneurysm formation in patients with a history of intracranial aneurysms. Lancet Neurol. 2014;13(1):59-66.

Zhang B et al. Application of Radiomics in Intracranial Aneurysm. Front Neurol. 2021;12:756350.