MR纹理分析鉴别诊断脑胶质母细胞瘤和单发转移瘤的价值

发布时间：2018-01-28 23:56

本文关键词： 胶质母细胞瘤肿瘤转移磁共振成像纹理分析　出处：《中国医学影像技术》2017年12期 　论文类型：期刊论文

【摘要】：目的探讨MR纹理分析鉴别脑胶质母细胞瘤和单发转移瘤的价值。方法收集我院病理或临床证实为脑胶质母细胞瘤患者35例、脑单发转移瘤患者40例。所有患者均接受常规MR平扫及增强检查,包括T1FLAIR、T2WI、T2FLAIR、DWI及增强T1WI。采用MaZda软件于DWI及增强T1WI序列最大层面上手动勾画肿瘤ROI,提取病变的纹理特征。结果脑胶质母细胞瘤和单发转移瘤的直方图纹理参数中增强T1WI的偏度,灰度共生矩阵参数中增强T1WI自相关、均值,DWI自关性差异有统计学意义。基于这些纹理参数构建多变量Logistic回归分析显示该模型的ROC曲线下面积为0.834。结论纹理分析为鉴别脑胶质母细胞瘤和单发转移瘤,提供可靠、可以量化的客观依据。
[Abstract]:Objective to study the value of Mr texture analysis in differentiating glioblastoma from single metastatic tumor. Methods 35 cases of glioblastoma proved by pathology or clinic in our hospital were collected. There were 40 patients with single metastatic tumor of brain. All the patients underwent conventional Mr plain scan and enhanced Mr examination, including T1FLAIRI T2WIT2FLAIR. DWI and enhanced T1WI. MaZda software was used to describe tumor ROI manually on the maximum plane of DWI and enhanced T1WI sequence. Results in the histogram texture parameters of glioblastoma and single metastatic tumor, the deviation of T1WI was enhanced, and the autocorrelation and mean value of T1WI in gray level co-occurrence matrix. The difference of DWI self-correlation is statistically significant. Multivariate Logistic regression analysis based on these texture parameters shows that the area under the ROC curve of the model is 0.834. Conclusion texture analysis is useful for discriminating. Glioblastoma and single metastatic tumor. Provide reliable, quantifiable objective basis.
【作者单位】：皖南医学院第一附属医院医学影像中心;浙江大学附属第一医院放射科;
【基金】：皖南医学院中青年科研基金自然科学类资助项目(WK2016F24)
【分类号】：R445.2;R739.41
【正文快照】： 胶质母细胞瘤和脑转移瘤是颅内较常见的恶性肿瘤,当脑转移瘤为单发灶且缺乏原发肿瘤病史时,与高级别脑胶质细胞瘤的影像表现非常相似,鉴别诊断困难[1]。目前MR是诊断脑肿瘤的主要影像学方法,主要序列为常规MR平扫、DWI、MR灌注成像和动态对比增强MRI,但MR的鉴别诊断脑肿瘤主要

【相似文献】