基于感兴趣区域灰度统计量的颅脑CT图像检索

针对目前只限于特定任务的医学图像检测方法，根据颅脑CT图像的特点，给出了将感兴趣区域灰度统计量作为参数检索图像的方法.首先利用区域生长法提取CT图像中的全局感兴趣区域颅脑，采用分块的方法将全局区域划分成多个子区域，以获得局部灰度空间信息，使整幅图像的检索转变为对各子区域的识别.进一步提取每个子区域的灰度统计量平均值、均方差和三阶矩作为特征，检索时将示例图像特征与特征库中的特征按照欧几里得距离进行相似性匹配.实验结果表明，该方法可查找出在相同位置具有相似病理特征的颅脑CT图像.

Brain CT image retrieval based on gray statistic features of interested region

An image retrieval algorithm taking gray statistics of interested region as parameter was proposed based on the features of brain CT image.The whole interested brain region was extracted from the CT image of by region growing algorithm.In order to get local gray spatial information,the interested image was segmented into several sub-regions.The retrieval of the whole region was changed into the recognition of each sub-region.The average,mean square error and three-order matrix of gray statistics of each sub...