| 李佳蔚,张源,刘文亚.基于门静脉期CT图像3D纹理分析评估肝泡型棘球蚴病生物学活性[J].中国医学影像技术,2023,39(11):1670~1674 |
| 基于门静脉期CT图像3D纹理分析评估肝泡型棘球蚴病生物学活性 |
| 3D texture analysis based on portal vein phase CT images for evaluating biological activity of hepatic alveolar echinococcosis |
| 投稿时间：2023-07-07 修订日期：2023-10-13 |
| DOI：10.13929/j.issn.1003-3289.2023.11.016 |
| 中文关键词: 棘球蚴病,肝 体层摄影术,X线计算机 生物学活性 纹理分析 |
| 英文关键词:echinococcosis, hepatic tomography, X-ray computed biological activity texture analysis |
| 基金项目:国家自然科学基金(81974263)。 |
|
| 摘要点击次数: 1693 |
| 全文下载次数: 551 |
| 中文摘要: |
| 目的 [JP3]观察门静脉期CT图像3D纹理分析评估肝泡型棘球蚴病(HAE)生物学活性的价值。方法 回顾性分析110例经术后病理确诊HAE患者的上腹部CT及PET/CT资料,依据PET/CT结果将其分为活性组(n=55)及无活性组(n=55)。采用MaZda软件于门静脉期CT图中逐层勾画病灶ROI,获得感兴趣容积(VOI);提取其直方图3D纹理特征参数并进行组间比较,将差异有统计学意义者纳入logistic回归分析行变量间共线性检验,以无共线性变量构建联合模型。绘制受试者工作特征曲线,计算曲线下面积(AUC),评估单一纹理特征及联合模型评估HAE病灶生物学活性的效能。结果 3D纹理特征参数中,组间均值、变异度、偏度、峰度、第50百分位数、第90百分位数和第99百分位数差异均有统计学意义(P均＜0.05);其中,均值、变异度、偏度、峰度及第99百分位数无共线性,单独以之判断HAE病灶生物学活性的AUC为0.614~0.628,而联合模型的AUC为0.714。结论 门静脉期CT图像3D纹理分析对于判断HAE病灶生物学活性具有一定价值。 |
| 英文摘要: |
| Objective To explore the value of 3D texture analysis based on portal vein phase CT images for evaluating biological activity of hepatic alveolar echinococcosis (HAE). Methods Data of upper abdominal CT and PET/CT of 110 patients with HAE diagnosed by postoperative pathology were retrospectively analyzed. The patients were divided into active group (n=55) or inactive group (n=55) according to PET/CT findings. MaZda software was used to delineate lesion ROI layer by layer on portal vein stage CT images, then lesion volume of interest (VOI) was generated, and its histogram 3D texture feature parameters were extracted and compared between groups, and parameters with significant differences were included in logistic regression analysis. Collinearity test was then carried out among variables, and a combined model was constructed with non-collinearity variables. Receiver operating characteristic curve was drawn, and area under the curve (AUC) was calculated to evaluate the efficacy of model of each single texture feature parameter and combined model for evaluating biological activity of HAE lesions. Results Significant differences of the mean, variability, skewness, kurtosis, 50th percentile, 90th percentile and 99th percentile of parameters of 3D texture feature were found between groups (all P＜0.05), among which no collinearity was found in the mean, variance, skewness, kurtosis and 99th percentile (all P＜0.05), with AUC for evaluating biological activity of HAE lesions of 0.614—0.628, while AUC of the combined model was 0.714. Conclusion 3D texture analysis based on portal vein CT images had certain value for evaluating biological activity of HAE lesions. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
