表观扩散系数直方图预测子宫动脉栓塞术治疗子宫腺肌病疗效的临床价值
Value of the Apparent Diffusion Coefficient Histogram in Outcome Prediction of Uterine Artery Embolization for Adenomyosis
朱晓青 1金文韬2
作者信息
- 1. 上海交通大学医学院附属第九人民医院 放射科,上海 200011
- 2. 复旦大学附属妇产科医院 放射科,上海 200011
- 折叠
摘要
目的 探讨表观扩散系数(Apparent Diffusion Coefficient,ADC)直方图数据在预测子宫动脉栓塞术(Uterine Artery Embolization,UAE)治疗子宫腺肌病疗效的临床价值.方法 回顾性分析90例于我院行UAE治疗的子宫腺肌病患者,收集术前MR图像,在弥散加权成像图像上测量子宫腺肌病病灶的直方图参数,包括10%、25%、50%、75%和90%的百分位数、最大、最小以及平均ADC值、峰度和偏度.根据术后病灶坏死情况进行分组,比较完全坏死组(66例)与不完全坏死组(24例)的参数,并通过受试者工作特征(Receiver Operating Characteristic,ROC)曲线评价各参数的预测效能.通过Logistic回归分析筛选变量,构建疗效预测模型并使用ROC曲线检验模型效能.结果 完全坏死组的10%、25%ADC值、平均ADC、最小ADC值均显著低于不完全坏死组(P<0.05).绘制ADC直方图各个参数的ROC曲线,曲线下面积(Area Under Curve,AUC)最高的参数为10%ADC值,其AUC为0.806,界值为0.963×10-3 mm2/s,敏感度为87.9%,特异性为70.8%.其次为最小ADC值以及25%ADC,AUC分别为0.790和0.771.将10%ADC、25%ADC以及最小ADC值的ROC曲线进行两两比较,三者之间的诊断效能差异均无统计学意义,P值分别为0.696(10%ADC与最小ADC值)、0.090(10%ADC与25%ADC)和0.704(最小ADC与25%ADC).通过Logistic回归分析将10%ADC和90%ADC纳入预测模型,其AUC值为0.826.结论 ADC直方图参数在预测UAE治疗子宫腺肌病疗效方面有潜在临床价值.
Abstract
Objective To explore the clinical value of the apparent diffusion coefficient(ADC)histogram analysis in the prediction of treatment response of uterine artery embolization(UAE)for adenomyosis.Methods A total of 90 patients with adenomyosis who underwent UAE treatment in our hospital were analyzed retrospectively.Preoperative MR images were collected and the histogram parameters of adenomyosis lesions were measured on diffusion-weighted imaging images.ADC histogram parameters including 10%,25%,50%,75%,and 90%percentiles,maximum,minimum and average ADC values,kurtosis and skewness.The patients were divided into complete necrosis group(n=66)and incomplete necrosis group(n=24)according to the degree of necrosis after surgery,and the parameters were compared between the two groups.The receiver operating characteristic(ROC)curve was used to evaluate performance of ADC histogram parameters in the prediction of the UAE for adenomyosis.The variables were screened by Logistic regression analysis,the curative effect prediction model was constructed and the model efficiency was tested by ROC curve.Results The 10%and 25%ADC percentiles,minimum and mean ADC value of complete necrosis group were significantly lower than that of incomplete necrosis group(P<0.05).The ROC curve of each parameter of ADC histogram was drawn.The largest area under curve(AUC)was 10%ADC percentiles.The AUC,threshold,sensitivity and specificity were 0.806,0.963×10-3 mm2/s,87.9%and 70.8%,respectively.Followed by the minimum ADC value and 25%ADC,AUC were 0.790 and 0.771,respectively.Compared the ROC curves of 10%ADC,25%ADC and the minimum ADC value,there was no statistical significance among them,and the P values were 0.696(10%ADC and minimum ADC value),0.090(10%ADC and 25%ADC value)and 0.704(minimum ADC and 25%ADC value),respectively.10%ADC and 90%ADC were included in the prediction model by Logistic regression analysis,and the AUC value for this model was 0.826.Conclusion ADC histogram parameters have potential clinical value in predicting the curative effect of UAE on adenomyosis.
关键词
子宫腺肌病/子宫动脉栓塞术/磁共振成像/表观扩散系数Key words
adenomyosis/uterine artery embolization/magnetic resonance image/apparent diffusion coefficient引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据