中文摘要：

提出一种基于CT图像的颅脑病变计算机自动检出方法。从正常人颅脑CT图片中手动分割出脑灰质、脑白质和脑脊液样本图片作为训练样本,然后采用模糊C均值聚类（FCM）算法进行模式分类,得到各聚类中心。对于待检测的CT图片,首先采用计算机自动分割算法进行分割,在尽量保证分割准确的前提下允许一定的过分割,将分割后的子图进行延拓和特征提取后与聚类中心进行距离计算,判断其是否属于三类正常样本,否则属于病变。通过实验验证了该算法的有效性。

英文摘要：

A new method of automatic detection of brain lesion based on wavelet feature vector of CT images has been proposed in the present paper.Firstly,we created training samples by manually segmenting normal CT images into gray matter,white matter and cerebrospinal fluid sub images.Then,we obtained the cluster centers using FCM clustering algorithm.When detecting lesions,the CT images to be detected was automatically segmented into sub images,with a certain degree of over-segmenting allowed under the premise of ensuring accuracy as much as possible.Then we extended these sub images and extracted the features to compute the distances with the cluster centers and to determine whether they belonged to the three kinds of normal samples,or,otherwise,belonged to lesions.The proposed method was verified by experiments.