一种CT医学图像分割新算法

A Novel Algorithm of CT Medical Image Segmentation

针对临床辅助诊断的需要,提出了一种结合传统分水岭算法和新型核聚类算法的CT医学图像分割新算法。首先,通过分水岭变换,CT图被分割成不同的小区域。然后,根据改进的KFCM算法,利用Mercer核将各个小区域的平均灰度值映射到高维特征空间,使得原来在分水岭算法分割图中未显示出来的特征显现出来。通过此方法,相较于传统核聚类(KFCM)算法,我们可实现更准确的聚类,并有效解决分水岭算法分割CT医学图像的过分割问题,因此能取得更好的分割效果。实验结果显示,本文方法能够很好分割腹腔CT图,获得更清晰的分割图像。

A novel method of CT image segmentation is presented for the need of computer-aided clinical diagnosis,which combines the conventional watershed with newer kernel-clustering algorithm.A CT image was first segmented into different small areas using watershed transform,and then,with an improved kernel-clustering algorithm,the mercer-kernel of WKFCM was used to map the average gray value of each small area of the segmented image into a high-dimensional feature space,making features not displayed in the conventional algorithm outstanding.In this way,more accurate clustering was achieved as compared with the conventional kernel fuzzy c-means（KFCM） clustering algorithm,and the problem of over-segmentation effectively solved which had dogged the watershed transform in segmenting CT images.Experimental results in segmenting abdominal CT images demonstrated satisfactory improvement of image quality.