| 赵正宇,单奔,韩雷,葛芳,李绍东.基于T2WI纹理分析预测高级别胶质瘤术后复发[J].中国医学影像技术,2023,39(11):1639~1643 |
| 基于T2WI纹理分析预测高级别胶质瘤术后复发 |
| Texture analysis based on T2WI for predicting recurrence of high-grade glioma after surgery |
| 投稿时间：2023-07-07 修订日期：2023-08-25 |
| DOI：10.13929/j.issn.1003-3289.2023.11.010 |
| 中文关键词: 胶质瘤 纹理分析 磁共振成像 |
| 英文关键词:glioma texture analysis magnetic resonance imaging |
| 基金项目: |
|
| 摘要点击次数: 2183 |
| 全文下载次数: 549 |
| 中文摘要: |
| 目的 观察基于T2WI纹理分析预测高级别胶质瘤(HGG)术后复发的价值。方法 回顾性分析71例术前接受MR检查的HGG患者,根据术后有无HGG复发分为复发组(n=45)和未复发组(n=26), 比较组间MRI肿瘤形态参数的差异;于T2WI中提取瘤体、瘤周水肿区及瘤体+瘤周水肿区的纹理特征,分别基于形态特征和纹理特征构建支持向量机(SVM)和随机森林(RF)模型;绘制受试者工作特征曲线,以曲线下面积(AUC)评估各模型的预测效能。结果 组间肿瘤位置、有无囊变、有无子病灶、实质/瘤周DWI信号及强化程度差异均有统计学意义(P均＜0.05)。分别于瘤体、瘤周水肿区和瘤体+瘤周水肿区筛选出12、13和13个最佳纹理特征,以之构建的SVM形态模型预测HGG复发效能 (AUC=0.76)高于RF形态模型(AUC=0.68),SVM纹理模型预测效能最佳(AUC=0.83)。结论 基于T2WI纹理分析能有效预测HGG术后复发,尤以SVM纹理模型的预测效能最佳。 |
| 英文摘要: |
| Objective To explore the value of texture analysis based on T2WI for predicting recurrence of high-grade glioma (HGG) after surgery. Methods Data of 71 HGG patients who underwent MR examination before surgery were retrospectively analyzed. The patients were divided into recurrent group (n=45) or non-recurrent group (n=26) according to postoperative recurrence or not. Preoperative MRI morphological findings of HGG were compared between groups. The texture characteristics of solid part of HGG, peritumoral edema area, as well as solid part of HGG+peritumoral edema area were extracted from T2WI, and then support vector machine (SVM) and random forest (RF) models were established based on morphological features and texture features, respectively. Receiver operating characteristic curves were drawn, and areas under the curves (AUC) were used to evaluate the prediction efficacy of each model. Results Significant differences of tumor location, cystic changes and sub lesions, parenchyma or peritumoral DWI signals and enhancement degrees were found between groups (all P＜0.05). There were 12, 13 and 13 optimal texture features being selected from solid part of HGG, peritumoral edema area and solid part of HGG+peritumoral edema area, respectively. SVMmorphology model (AUC=0.76) had higher efficacy for predicting recurrence of HGG than RFmorphology model (AUC=0.68), while SVMtexture model had the best performance (AUC=0.83). Conclusion Texture analysis based on T2WI was effective for predicting recurrence of HGG after surgery, and SVMtexture model had the best performance. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
