Development and validation of sonographic feature-based prediction models for ultrasound-guided HIFU ablation of uterine fibroids.

Abstract

Purpose: To develop a nomogram for predicting the efficacy of high-intensity focused ultrasound (HIFU) ablation in patients with uterine fibroids.

Methods: A retrospective analysis was conducted on 153 patients (39.84 ± 6.37 years old, 224 fibroids) who underwent HIFU. Patients were randomly divided into training group (n = 162) and validation group (n = 62) in a 7:3 ratio. Pre-HIFU ultrasonic features of fibroids were observed. An effective treatment was defined as an ablation rate of ≥70%. In the training group, Lasso regression was used to identify independent predictive factors, and a nomogram model was developed. Internal validation was conducted using the Bootstrap method, and model discrimination was evaluated with ROC curves. Model accuracy was assessed using calibration curves, and clinical benefits were analyzed using decision curve analysis (DCA).

Results: The Lasso regression model identified uterine fibroid type, peripheral blood flow grading, fibroid location, and maximum fibroid diameter as potential predictive factors. The nomogram showed good predictive performance with AUCs of 0.819 (95% CI 0.789-0.822) in the training set and 0.900 (95% CI 0.840-0.901) in the validation set. Its sensitivity and specificity were 0.750/0.732 in the training set and 0.759/0.818 in the validation set, respectively. The calibration curve showed good agreement between observed and predicted outcomes, and DCA confirmed the clinical utility and broad applicability of the nomogram model.

Conclusion: The prediction model comprising ultrasonic characteristics of uterine fibroids had good performance in patients undergoing HIFU ablation and hence can improve clinical decision-making.