摘要
目的采用反映血管新生状态的指标经模糊C均值聚类对星形细胞肿瘤病理学分级进行探讨。方法采用含有正常成人脑组织、弥漫性星形细胞瘤(WHOⅡ级)、间变性星形细胞瘤(WHOⅢ级)、胶质母细胞瘤(WHOⅣ级)及阳性对照组织的168点矩阵的组织芯片,通过免疫组织化学SABC双标法标记内皮细胞和血管内皮生长因子,以Image-Pro Plus 5.1中文版图像分析软件对染色结果及血管内皮生长因子阳性单位、微血管密度及微血管平均周长等指标进行测定。采用单因素分析方法筛选与星形细胞肿瘤病理级别相关的参数,以矩阵实验室数学软件提供的模糊C均值聚类函数参数作为聚类对象,将不同的组织切片参数值进行模糊C均值聚类,所得聚类值分别赋值为星形细胞肿瘤病理分级值。结果(1)在不同病理分级组之间,星形细胞肿瘤血管内皮生长因子阳性单位差异具有统计学意义(P=0.000),各病理分级组间两两比较差异亦有统计学意义(均P<0.05)。(2)在不同病理分级组之间,星形细胞肿瘤微血管密度值差异有统计学意义(P=0.000),两两比较差异亦有统计学意义(均P=0.000)。(3)星形细胞肿瘤微血管平均周长,Ⅱ级组与Ⅲ级组、Ⅱ级组与Ⅳ级组比较差异有统计学意义(均P=0.000),而Ⅲ级组与Ⅳ级组之间差异无统计学意义(P=1.000)。(4)与WHO病理分级相比,模糊C均值聚类产生的星形细胞肿瘤病理分级值对Ⅱ、Ⅲ、Ⅳ级等级别的诊断符合率分别为85.71%、48.39%和78.95%,总体正确率达68.46%。结论星形细胞肿瘤血管内皮生长因子阳性单位、微血管密度和微血管平均周长等项指标的模糊C均值聚类值与星形细胞肿瘤病理分级值比较符合,可应用模糊C均值聚类法对星形细胞肿瘤的病理分级进行辅助推测。
Objective To investigate the fuzzy C-means （FCM） clustering method for the reflection of angiogenesis status in the pathological grading of astrocytoma. Methods A tissue microarray with 168 points including normal adult brain tissue, diffuse astrocytoma （WHO grade Ⅱ ）, anaplastic astrocytoma （WHO grade Ⅲ）, glioblastoma （WHO grade Ⅳ） and positive control tissues was employed. Double-labelling technique of immunohistochemistry staining was applied to mark endothelial cell and vascular endothelial growth foctor （VEGF）. Image-Pro Plus （5.1 Chinese version） was used to measure the staining results and the indexes of VEGF positive unit （PU）, micro-blood vessel （MBV） density, mean MBV perimeter. Screened the parameters related to the grade of astrocytoma by single factor analysis. On basis of these parameters from different tissue sections, we used FCM clustering method from the mathematical software in MATLAB to get the value of the FCM cluster and to compare it with the pathological grades of astrocytoma. Results 1） The differences of VEGF PU value in different grade groups of astrocytoma were significant （P=0.000）, and the difference between each pathological grade group was also significant （P〈 0.05, for all）. 2） The differences of MBV density in astrocytoma in different pathological grade groups were significant （P= 0.000）, and the difference between each grade group was also significant （P=0.000, for all）. 3） The differences between grade Ⅱ group and grade Ⅲgroup, and grade Ⅱgroup and grade Ⅳgroup in mean MBV perimeter were significant （P= 0.000, for all）, but no significant difference was seen between grade Ⅲ group and grade Ⅳgroup （P= 1.000）. 4） In comparison with WHO pathological grade, the diagnostic coincidence rates of astrocytoma pathological grade Ⅱ, Ⅲ, and Ⅳ idemtified by FCM clustering method were 85.71% , 48.39%, and 78.95% respectively, and the total correct rate was 68.46%. Conclusion The indexes of VEGF PU,MBV density and mean MBV perimeter in FCM clustering method relatively coincide with the pathological grades of astrocytoma. FCM clustering method can be an accessory inferential method for the pathological grading of astrocytoma.
出处
《中国现代神经疾病杂志》
CAS
2007年第3期235-239,共5页
Chinese Journal of Contemporary Neurology and Neurosurgery
基金
国家自然科学基金资助项目(项目编号:30370552)
