Value of multi-modal MRI in predicting the effect of high-intensity focused ultrasound for uterine fibroids.

Abstract

Objective: To explore the value of advanced MRI techniques in predicting the effect of high-intensity focused ultrasound (HIFU) on uterine fibroids and to establish a prediction model.

Methods: This study enrolled 40 patients with 52 uterine fibroids who underwent HIFU at our hospital. All patients were scanned using multiple MRI sequences before HIFU therapy, including non-contrast enhanced MRI, contrast-enhanced MRI, diffusion weighted imaging (DWI), arterial spin labeling (ASL), and T1 mapping. MRI sequences that could predict the HIFU effect were identified, and the predictive performance was evaluated using ROC curves. Univariate and multivariate logistic analyses were employed to investigate independent predictors and establish a prediction model. In addition, we assessed and verified the performance of the model.

Results: Hyperintense on T2 weighted imaging (T2WI), large apparent diffusion coefficient (ADC) values derived from DWI, and high perfusion index derived from ASL were associated with a poor HIFU effect. Univariate and multivariate logistic analyses suggested that uterine position, ADC value and perfusion index were independent predictors for establishing the prediction model. The AUC of the prediction model was 0.939. Both the Hosmer-Lemeshow test and the calibration curve indicated good calibration. The decision curve analysis (DCA) curve showed good clinical benefits, and the leave-one-out cross-validation (LOOCV) revealed that the model had good predictive performance and generalization ability.

Conclusion: ADC values and perfusion index are predictors of the effect of HIFU on uterine fibroids. The prediction model including uterine position, ADC value, and perfusion index as predictors has good predictive performance.