Magnetic resonance guided high intensity focused ultrasound (MR-HIFU) is a noninvasive therapy for the ablation of symptomatic uterine fibroid. The non-perfused volume (NPV),which reflect an immediate ablation efficiency, is one of the most important factor affecting clinical efficacy[1]. Kim reported that for symptomatic uterine fibroids ,a higher K^trans value which derived from DCE-MRI with averaging each AIF at baseline suggested a poor ablation efficacy of MR-HIFU[2]. Reference region model(RR model) is a method that compares the tissues of interest curve shape to that of a reference region ,with no need for direct AIF measurement[3].The aim of this study is to investigate whether the parameters of quantitative DCE-MRI could be a predictor of ablation efficiency in MR-HIFU of uterine fibroids based on RR model and 3D entire-tumor histogram analysis. This prospective study was approved by our institutional review board. All MR examinations were performed on a 1.5-Tesla MR unit (Achieva). 24 eligible females (43.8years;37~54 years) underwent DCE-MRI with intravenous administration of Magnevist (0.1 mmol/kg) during screening procedure. After therapy with volumetric MR-HIFU ablation, same DCE-MRI was performed immediately to determine the NPV. RR model was applied to acquire quantitive parameters of DCE-MRI (RR-K^trans and RR-K_ep ) by using the commercial analysis software (Omni-Kinetics 2.0). ROI measurement was based on slice-by-slice 3D entire tumor and 2D center slice, repectively. The Wilcoxon rank-sum test was used to analyze the difference of each histogram metrics between the two groups（NPV_high, NPV ratio higher than 60%; NPV_low, NPV ratio lower than 60%). The relationship between NPV ratio and parameters of DCE-MRI was determined by using spearman correlation coefficient.24 patients were divided into NPV_high (n=11) and NPV_low (n=13). Seven 3D entire-tumor histogram metrics of RR-K^trans showed statistical significance between the two groups (Table 1), while all metrics of RR-k_ep displayed no statistical significance. Similarly to the 2D mean value, no statistical significance was observed between the two groups. Figure 1 showed the z-scored statistical results of entire-tumor and 2D mean values in discriminating NPV_high and NPV_low. The correlation coefficient between NPV ratio and RR-K^trans is -0.6. Figure 2 showed the typical histogram distributions of a NPV_high case and a NPV_low case, respectively. Among the seven 3D metrics of RR-K^trans, ROC analysis revealed that mean, skewness, and median have relatively larger AUC(>0.8) in differentiating high and low NPV ratio (Table 2).DCE-MRI could provide information about blood perfusion and microvascular permeability. To estimate the kinetic parameters of DCE-MRI, accurate and reproducible AIF is very important[4]. However,accurate AIF of internal iliac artery was difficult to acquire[5]. As an substituted approach, RR model has been developed for analyzing DCE-MRI that do not require direct measurement of the AIF[6]. In addition, blood supply of uterine fibroids and gluteus maximus are all derived from the internal iliac artery. Therefore, the gluteus maximus as a reference region was used to process DCE-MRI analysis based on RR model in this study. NPV ratio was an important factor for assessing the therapy efficacy. In current study, NPV ratio of 60% as cut-off point for grouping was inconsistent with NPV ratio of 80% in a previous study [7], which attributed to individual differences of recruited cases. Since the NPV ratio derived from entire-tumor, we used entire-tumor metrics of DCE-MRI and compared with those of 2D center slice of fibroids. All the 3D entire-tumor metrics of RR-K^trans showed statistical significance in discriminating NPV_high from NPV_low. It suggested that entire-tumor histogram analysis might play better role than traditional 2D measurement in the evaluation of MR-HIFU ablation efficacy. Meanwhile, histogram metrics revealed comparative performance with mean in discriminating the two groups. As shown in Table 2, mean obtained a sensitivity of 92.3% and specificity of 72.7%; while median performed a good specificity of 100% though with a slighter low sensitivity of 61.5%. Moreover, skewness showed the same AUC as mean. It was higher in NPV_high group than in NPV_low group, which meaned RR-K^trans towards zero side and blood perfusion was relatively lower. Our findings showed that low perfusion fibroids with low RR-K^trans may get higher NPV ratio in MR-HIFU. Blood temperature of ablated fibroids is lower than that of the surrounding tissue. The capillary can take away the heat accumulation within tissue, leading to local tissue temperature insufficient to effective ablation. So the ablation is difficult to achieve the desired temperature in high blood perfusion tissue. In conclusion, our results indicated that the 3D histogram analysis may produce more comprehensive and accurate features of fibroid perfusion, and RR-K^trans might be sensitive for predicting the immediate ablation efficiency of MR-HIFU.