Beyond Skull Density Ratio for Magnetic Resonance‑Guided Focused Ultrasound: A Literature Review of the Alternatives.

BACKGROUND AND

OBJECTIVES

Motivated by growing literature suggesting that skull density ratio (SDR) has limitations in patient selection and outcome prediction for magnetic resonance‑guided focused ultrasound treatments, this study sought to systematically review imaging-based alternatives to SDR that have been explored in relation to clinical and technical outcomes.

METHODS

This review followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses reporting guidelines and was registered in the International Prospective Register of Systematic Reviews (CRD420251081059). We searched MEDLINE, Embase, and Scopus from inception to February 2025 for studies evaluating imaging-derived alternatives to SDR in human magnetic resonance‑guided focused ultrasound procedures. Data were extracted independently by 2 reviewers and studies were grouped into 4 categories: skull geometric factors, histogram-based SDR analysis, patient-specific multivariate modeling, and advanced imaging processing. Each category was also assessed for clinical implementation feasibility based on imaging processing complexity, required expertise, and scalability.

RESULTS

Of 1684 screened studies, 23 met the inclusion criteria. Skull geometric factors (n = 9), particularly skull thickness and volume, were the most commonly studied and showed consistent associations with both thermal and clinical outcomes. Histogram-based SDR metrics (eg, skewness) occasionally had stronger correlations with outcomes than mean SDR. Multivariate models and advanced imaging showed strong technical correlations but comparatively lower clinical feasibility due to complexity and computing demands.

CONCLUSION

Although SDR remains the approved screening metric, our review suggests potential imaging-based alternatives to SDR. Specifically, readily available and implementable metrics such as skull geometric features show promising correlations with clinical and technical outcomes. Other methods, such as multivariate modeling, show promise but will need more time to become validated, accessible, and widely implemented.