LAUSR

Abstract Objectives: The objective of this study is to assess the association between texture parameter analysis derived from T2-weighted images and efficiency of magnetic resonance-guided focussed ultrasound (MRgFUS) ablation for the treatment of uterine fibroids. Materials and methods: Fifty-five women treated by MRgFUS were included in this retrospective analysis. Texture parameters were calculated using three 2D regions of interest placed on three consecutive slices on the same sagittal 3D T2-weighted images obtained at the beginning of MRgFUS ablation. Using uni- and multi-variate linear regression, texture parameters, fibroids/muscular T2W ratio (T2Wr), Funaki type, and fibroid depth were correlated with ablation efficiency, defined as the ratio of non-perfused volume (NPV) on post-treatment contrast-enhanced MRI by total volume of treatment-cell sizes used. Inter-rater reproducibility for texture analysis was assessed using variation coefficients. Results: The mean total treatment cell volume was 49.5 (±30) ml, corresponding to a mean NPV of 57.2 (±57) ml (28%). The mean ablation efficiency was 1.14 (±0.7), with a range of 0.03–3.6. In addition to fibroid/muscular T2Wr, seven of the 14 texture parameters were significantly correlated with ablation efficiency: mean signal intensity (p = .047); Skewness (p = .03); Kurtosis (p = .015); mean uniformity (p = .052); mean sum of square (p = .045); mean sum entropy (p = .021) and mean entropy (p = .051). In multivariate linear regression, fibroid/muscular T2Wr and sum of entropy were associated with ablation efficiency. The inter-rater coefficient of variation for sum entropy was 2.6%. Conclusions: Uterine fibroid texture parameters provide complementary information to T2Wr, and are associated with MRgFUS efficiency. Key points Mean sum entropy is negatively correlated with MRgFUS efficiency (ρ = −0.307, p = .021). Fibroids/muscular T2-weighted ratio and entropy are associated with MRgFUS efficiency. Texture parameters are better predictors of MRgFUS efficiency than Funaki type. Fibroid MR texture analysis can improve patient selection for MRgFUS.