正常人脑^(18)F-FDG PET显像的解剖功能区自动提取法分析

Automatic extraction analysis of the anatomical functional area for normal brain ^(18)F-FDG PET imaging

应用自行设计的脑解剖功能区自动提取软件研究了正常人脑18F-FDG代谢显像的灰度分布及其与剂量、血糖水平以及年龄等因素的关系。按照标准的Talairach脑坐标图谱,将脑18F-FDG代谢显像经过位移和形变配准到标准空间中去,提取出各解剖功能区象素,计算它们的平均灰度值,同时以灰度值为应变量,年龄、血糖以及剂量为自变量,按多元逐步回归统计方法,找出与灰度值相关的脑解剖区。结果显示:经过图像配准,额叶、顶叶、枕叶、颞叶和小脑以及丘脑、海马、第三脑室、第四脑室等区域可以进行成功提取,各功能区灰度值与小脑灰度值之比分布在1.01±0.15之间,多元逐步回归分析可见除尾状核和海马区域,另外提取的脑功能区的灰度值高低均与年龄大小成负相关,而所有区域的灰度与血糖和剂量均无相关性。表明自行设计的脑解剖功能区自动提取方法能方便地提取脑解剖功能区,并计算其各项参数作进一步的统计分析,但是解剖功能区的提取区域范围尚需更加细致化,才能准确研究认知和精神疾病领域的变化。

Using self-designed automatic extraction software of brain functional area, the grey scale distribution of 18F-FDG imaging and the relationship between the 18F-FDG accumulation of brain anatomic function area and the 18F-FDG injected dose, the level of glucose, the age, etc., were studied. According to the Talairach coordinate system, after rotation, drift and plastic deformation, the 18F-FDG PET imaging was registered into the Talairach coordinate atlas, and then the average gray value scale ratios between individual brain anatomic functional area and whole brain area was calculated. Further more the statistics of the relationship between the I8F-FDG accumulation of every brain anatomic function area and the I8F-FDG injected dose, the level of glucose and the age were tested by using multiple stepwise regression model. After images' registration, smoothing and extraction, main cerebral cortex of the 18F-FDG PET brain imaging can be successfully localized and extracted, such as frontal lobe, parietal lobe, occipital lobe, temporal lobe, cerebellum, brain ventricle, thalamus and hippocampus. The average ratios to the inner reference of every brain anatomic functional area were 1.01 ± 0.15. By multiple stepwise regresion with the exception of thalamus and hippocampus, the grey scale of all the brain functional area was negatively correlated to the ages, but with no correlation to blood sugar and dose in all areas. To the I8F-FDG PET imaging, the brain functional area extraction program could automatically delineate most of the cerebral cortical area, and also successfully reflect the brain blood and metabolic study, but extraction of the more detailed area needs further investigation.