【摘要】 提出了一套基于图像内容的颅脑病变计算机自动分析新方法。首先将图像分割为固定的感兴趣区域，采用统计量作为描述参数，然后用特征值通过决策树分类分析,提取出隐藏在其中的判定图像是否正常的规则。实验证明该方法对于第3脑室上部层面到大脑皮质上部层面CT图像的诊断正确率较高。更多还原

【Abstract】 In this paper, a new method to do automatic analysis of brain pathology based on CT image content is proposed. Image is segmented into fixed regions of interest and statistics are taken as description parameters. The feature parameters are analyzed by decision tree in order to find the hidden rules for diagnosing. Experiment result proves that this method gets high accuracy rate for the images from the third ventricles of brain upper layer to cerebral cortex upper layer.更多还原