Abstract: In order to improve the application effect, stability and quality of K-means clustering algorithm in medical CT image segmentation and reduce the running time of the program, K-clustering algorithm program was optimized with Matlab language in this study, and the CT image segmentation experiments were also done on the original medical CT images data set with the optimized K-clustering algorithm program and the K-means function of the MATLAB R2012a Statistics Toolbox. It is analyzed the running time differences of the two kinds of program with single factor variance analysis method, and the stability and quality of the images segmentation results were observed. The results show that the improved K-means clustering algorithm programming had higher stability and quality of segmentation results, etc. Under the environment of ordinary Windows operation system and hardware of personal computer configuration, the segmenting time was about one second, significantly lower than that of the original segmentation procedures, thus the feeling of waiting was eliminated, the user’s comfort and efficiency were improved, and the foundation was laid for the recognition processing of the image.

Key words: image segmentation, K-means clustering, computer tomography image