Entropy of T2-weighted imaging combined with apparent diffusion coefficient in prediction of uterine leiomyoma volume response after uterine artery embolization.

RATIONALE AND

OBJECTIVES

To determine the potential value of entropy of T2-weighted imaging combined with apparent diffusion coefficient (ADC) before uterine artery embolization (UAE) for prediction of uterine leiomyoma volume reduction (VR) after UAE.

MATERIALS AND

METHODS

In this prospective study, 11 patients with uterine leiomyomas who underwent pelvic magnetic resonance imaging including diffusion-weighted imaging before and 6 months after UAE were included. A total number of 16 leiomyomas larger than 2 cm in diameter were evaluated. The volume of each leiomyoma before and after UAE was determined, and the percentage change in volume was calculated. Entropy of T2-weighted imaging and ADC before UAE were assessed. Pearson correction coefficients were calculated between leiomyoma VR after UAE and age, leiomyoma volume, ADC, and entropy, respectively. Multiple regression analysis was performed to investigate the parameters that determine the VR after UAE. Receiver operating characteristic curve analysis was used to determine the sensitivity and specificity of ADC, entropy and the combination of ADC and entropy for predicting volume response.

RESULTS

The mean leiomyoma VR was 58.9% (range 25.8%-95.0%) in the 6-month follow-up. The mean ADC of leiomyomas was 1.37 × 10(-3) mm(2)/s (range 1.05 × 10(-3)-2.32 × 10(-3) mm(2)/s) and the mean entropy of T2-weighted imaging was 5.36 (range 4.62-5.91) before UAE. ADC and entropy were significantly correlated with leiomyoma VR, respectively (r = 0.61, P = .012; r = 0.73, P = .001). On multiple regression analysis, a combination of ADC and entropy constituted the best model for determining leiomyoma VR using Akaike information criterion. For predicting ≥50% VR, the optimal cutoff value of ADC was 1.39 × 10(-3) mm(2)/s (sensitivity 45.5%, specificity 80.0%) and the optimal cutoff value of entropy was 5.15 (sensitivity 90.9%, specificity 60.0%). The combination of ADC and entropy (area under the curve [AUC] 0.86) provided better classification accuracy than ADC or entropy alone (AUC 0.69 and 0.82, respectively).

CONCLUSIONS

Pre-UAE entropy of T2-weighted imaging and ADC of leiomyomas were significantly correlated with the leiomyoma VR 6 months after embolization. Higher entropy and higher ADC may be related to greater leiomyoma VR after UAE. A combination of entropy and ADC may have predictive value for leiomyoma VR after UAE.