【摘要】 目的 分析基于MRI的影像组学联合影像学特征预测子宫肌瘤高强度聚焦超声（HIFU）术后消融率的价值。方法 搜集经影像学检查确诊的106例子宫肌瘤患者，根据HIFU消融效果分为高消融率组（消融率≥70%）、低消融率组（消融率<70%），分析子宫肌瘤HIFU消融治疗相关的影像学特征，勾画并筛选影像组学特征，构建并分析影像组学、影像学及影像组学-影像学模型预测HIFU术后消融率的价值。结果 高消融率组T2WI低信号占比、轻度强化占比更高（P<0.05），表观扩散系数平均值及最小值更低（P<0.05）。影像组学特征中筛查到2个灰度依赖矩阵、5个灰度共生矩阵、2个灰度游程矩阵、1个灰度大小区域矩阵、1个邻域灰度差矩阵、1个一阶特征。影像组学、影像学及影像组学-影像学模型预测HIFU低消融率的敏感度分别为79%、80%、83%，特异度分别为67%、71%、79%，曲线下面积分别为0.759、0.761、0.851。结论 基于MRI的影像组学联合影像学特征能够用于预测子宫肌瘤HIFU术后消融率，预测效能良好。更多还原

【Abstract】 Objective To analyze the value of MRI based radiomics combined with imaging features in predicting the postoperative ablation rate of uterine fibroids using high?intensity focused ultrasound （HIFU）.Methods 106 patients with uterine fibroids admitted were selected and divided into a high ablation rate group （ablation rate≥70%） and a low ab?lation rate group （ablation rate<70%） based on the HIFU ablation effect.The imaging features related to HIFU ablation treatment of uterine fibroids were analyzed,and the imaging features were delineated and screened.The value of ra?diomics,imaging,and radiomics?imaging fusion models for predicting HIFU ablation rate was constructed and analyzed.Results The high ablation rate group had a higher proportion of low signal and mild enhancement on T₂WI （P<0.05）,and lower average and minimum ADC values （P<0.05）.Two grayscale dependency matrices,five grayscale co?occurrence matrices,two grayscale run length matrices,one grayscale size region matrix,one neighborhood grayscale difference ma?trix,and one first?order feature were screened in the radiomics features.The sensitivity,specificity,and area under the curve of radiomics,radiomics,and radiomics imaging models for predicting low ablation rates of HIFU were 0.79,0.80,and 0.83,0.67,0.71,and 0.79,respectively.Conclusion The combination of MRI based radiomics and imaging fea?tures can be used to predict the ablation rate of uterine fibroids after HIFU surgery,with good predictive performance.更多还原