中文摘要：

针对脑肿瘤和脑出血的检出应用,提出一种新的将颅脑CT图像分割为白质、灰质和脑脊液3个区域的方法。首先用阈值法与数学形态学结合的方法提取出颅腔内脑部组织,然后利用灰度信息将其分割成脑脊液区域和其他区域。计算其他区域像素的几何矩及几何矩的方差,并结合像素的边界特性为每个像素构建特征向量。用M FCM算法对特征向量进行聚类将图像分割成白质和灰质,并用数学形态学对聚类结果进行滤波。实验结果验证了本文方法的先进性。

英文摘要：

Aimed at the detection of brain tumors and the cerebral hemorrhage,a new approach for tissue segmentation of brain CT data is proposed to label the white matter（WM）,the gray matter（GM） and the cerebrospinal fluid（CSF）.By using thresholding and mathematical morphology,the cerebrum is firstly extracted.After the cerebrum is approximately classified into CSF and the other part according to the discrepancy of their intensity,the geometric moment,its standard deviation,and the edge strength of each voxel of the other part are calculated and constructed as the eigenvector of the voxel.MFCM classifies the eigenvector into WM and GM,and the mathematical morphology modifies the clustering result.Experimental results demonstrate the good effect of the method.