| 贾颖,杜学松,陈君辉,张园园,方靖琴,张伟国.基于常规MRI的定量影像学特征用于胶质瘤分级诊断[J].中国医学影像技术,2018,34(8):1137~1142 |
| 基于常规MRI的定量影像学特征用于胶质瘤分级诊断 |
| Quantitative features extracted based on conventional MRI for grading diagnosis of glioma |
| 投稿时间：2018-01-29 修订日期：2018-05-29 |
| DOI：10.13929/j.1003-3289.201801178 |
| 中文关键词: 磁共振成像 影像学特征 定量参数 神经胶质瘤 肿瘤分级 |
| 英文关键词:Magnetic resonance imaging Image feature Quantitative parameters Glioma Neoplasm grading |
| 基金项目:重庆市科技研发基地建设计划项目（cstc2014gjhzl10002）。 |
|
| 摘要点击次数: 2988 |
| 全文下载次数: 981 |
| 中文摘要: |
| 目的 探讨常规MRI提取的定量影像学特征在胶质瘤分级诊断中的价值。方法 回顾性分析153例胶质瘤术前常规MRI资料，采用伦勃朗视觉感受图像（VASARI）特征集提取30个定量影像学特征（F1~F30），以Kappa检验验证其稳定性，筛选不同胶质瘤级别间有统计学差异的定量特征，并以Binary Logistic回归评价其用于胶质瘤分级诊断的价值。结果 153例中，WHO Ⅱ级胶质瘤68例，Ⅲ级34例，Ⅳ级51例。30个VASARI特征均有较高稳定性（Kappa值均>0.5），其中10个定量特征，包括脑功能区受累情况（F3）、强化程度（F4）、强化百分比（F5）、未强化百分比（F6）、坏死百分比（F7）、强化边缘厚度（F11）、强化边缘清晰度（F12）、扩散（F17）、室管膜侵犯（F19）及未强化区跨脑中线情况（F22）在不同级别胶质瘤中差异均有统计学意义（P均<0.01）。F3有助于鉴别高级别与低级别及WHO Ⅱ级与Ⅲ级胶质瘤（β=-0.683，OR=0.505，P=0.006）； F22对鉴别WHO Ⅲ级与Ⅳ级胶质瘤有较高诊价值（β=2.161，OR=8.682，P=0.008）。结论 依据VASARI特征集基于常规MRI提取的定量影像学特征稳定性较好，可用于术前分级诊断胶质瘤。 |
| 英文摘要: |
| Objective To explore the value of quantitative features derived from conventional MRI in grading diagnosis of glioma. Methods Pre-operative MRI data of 153 patients with glioma were analyzed retrospectively. The quantitative image features were extracted from conventional MRI using the visually accessible Rembrandt images (VASARI) features (F1-F30). The stability of 30 features were evaluated with Kappa test. The image features with significant differences of different grades of gliomas were obtained, and their grading diagnostic value were evaluated with Binary Logistic regression. Results In all the 153 patients, 68 patients were of WHO Ⅱ gliomas, 34 were of WHO Ⅲ and 51 were of WHO Ⅳ gliomas. The stability of all the features were high (all Kappa>0.5). Ten features, including eloquent brain (F3), enhancement quality (F4), proportion enhancing (F5), proportion non-enhancing tumor (F6), proportion necrosis (F7), thickness of enhancing margin (F11), definition of the enhancing margin (F12), diffusion (F17), ependymal invasion (F19), non-enhancing tumor crosses midline (F22) showed significant differences among different glioma grades (all P<0.01). F3 was helpful to differentiating high and low grade (regression coefficient=-0.467, odd ratio=0.627, P=0.005) gliomas and WHO Ⅲ and Ⅳ gliomas (β=-0.683, OR=0.505, P=0.006), while F22 had high value in differentiating WHO Ⅱ and Ⅲ gliomas (β=2.161, OR=8.682, P=0.008). Conclusion Quantitative image features extracted from conventional MRI are stable and helpful to pre-operative grading diagnosis of gliomas. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
