| 张冀,王鹰,李涛.基于因子分析的高级别脑胶质瘤MR灌注定量分析[J].中国医学影像技术,2017,33(1):119~123 |
| 基于因子分析的高级别脑胶质瘤MR灌注定量分析 |
| Study on MR perfusion quantification of high-grade brain glioma based on factor analysis |
| 投稿时间：2016-06-24 修订日期：2016-11-03 |
| DOI：10.13929/j.1003-3289.201606128 |
| 中文关键词: 胶质瘤 脑 磁共振成像 因子分析 高级别 |
| 英文关键词:Glioma Brain Magnetic resonance imaging Factor analysis High-grade |
| 基金项目:国家自然科学基金（81401474）、湖北省自然基金项目（2012FFB06809）。 |
|
| 摘要点击次数: 1831 |
| 全文下载次数: 825 |
| 中文摘要: |
| 目的 采用动态结构因子分析法（FADS）对5例高级别脑胶质瘤MR灌注图像进行定量分析，以准确提取脑胶质瘤MR灌注图像ROI的时间-信号曲线（TISCs）。方法 采用替代-近似算法对FADS模型求解，并分析从胶质瘤和正常组织区域提取的时间-信号曲线（TISCs）和因子图的特点。分别计算和比较胶质瘤与正常组织及病例间TISCs的相关系数。结果 采用FADS法从胶质瘤和正常组织中均提取到1条波峰向上的曲线（即升峰曲线）和2条波峰向下的曲线（分别为降峰曲线a和b）。胶质瘤与正常组织升峰曲线的相关系数平均值为0.75±0.10，明显低于胶质瘤患者间升峰曲线相关系数平均值0.84±0.05（P<0.05）。胶质瘤降峰曲线a与正常组织降峰曲线相关系数最大均值和胶质瘤降峰曲线b与正常组织降峰曲线相关系数最大均值差异有统计学意义（P<0.05）。胶质瘤患者因子图中瘤周区域主要对应升峰曲线，而胶质瘤区域主要对应两条降峰曲线。结论 采用FADS能自动提取到胶质瘤的TISCs，初步证明利用曲线的生理参数进行胶质瘤分级诊断具有可行性。 |
| 英文摘要: |
| Objective To quantitative analyze MR perfusion sequence images of 5 high-grade brain gliomas patients using factor analysis of dynamic structures (FADS), and to accurately extract the time-signal intensity curves (TISCs) of ROI from MR perfusion image sequence of brain glioma. Methods The FADS model was solved by replace-approximation method, and the characterization of TISCs and factor images from gliomas and normal tissue regions were analyzed. The correlation coefficients (CCs) of TSICs between gliomas and normal tissue and among the TISCs of patient were computed and compared, respectively. Results One crest-up (CU) curve and two crest-down curves (CD curve a and CD curve b) were extracted from the gliomas and normal tissue. The average value of CU curve CCs between the gliomas and the normal tissue were 0.75±0.10, which was obviously lower than those among glioma patients (0.84±0.05; P<0.05). Compared with the maximum average of CCs between the CD curve b of glioma and the CD curve of normal tissue, the maximum average of CCs between the CD curve a of glioma and the CD curve of normal tissue had obvious significance (P<0.05). In the factor images of the patients, the surrounding tissue of glioma was mainly corresponding to the CU curve, and the glioma was mainly corresponding to the CD curve. Conclusion The TISCs of glioma could be extracted automatically by FADS. It preliminarily demonstrats the feasibility of differentiation diagnosis on the grade of gliomas by using the physiological parameters of TISCs extracted by FADS. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
