Home    About Journal    Editorial Board    Instruction    Subscribe    Download    Messages Board    Contact Us    中文
  Apr. 5, 2025  
Quick Search
Chinese Journal of Biomedical Engineering  2018, Vol. 37 Issue (2): 138-145    DOI: 10.3969/j.issn.0258-8021.2018.02.002
Regular Papers Current Issue | Archive | Adv Search |
Cervical Image Segmentation Based on Modified k Means Algorithm and Gaussian Mixture Model
Liu Jun1*, Yu Tingting2, Shi Huijuan2
1Key Laboratory of Jiangxi Province for Image Processing and Pattern Recognition, Nanchang Hangkong University, Nanchang 330063, China
2College of Information Engineering, Nanchang Hangkong University, Nanchang 330063
Download: PDF (834 KB)   HTML (1 KB) 
Export: BibTeX | EndNote (RIS)      
Abstract  In order to segment the cervix region from colposcope images in the intelligent cervix cancer screening system, a kind of method based on modified k means algorithm and Gaussian mixture model was developed in this paper. Firstly, the data set to be classified was constructed according to the representative color of the cervix region and the distance to the image center point. Secondly, by recalculating the representative color of the cervix region, a dynamic regulation which enable the data set to be classified keep updating with the iteration was added in the k means algorithm, thus the k means algorithm can be applied to target images that obtained under different light environment. Lastly, the segmentation result was obtained by Gaussian mixture model that initialized by the result of the modified k means algorithm. 75 sets of cervical images that photographed under different conditions were used in the experiments. The results on these data showed that developed method gained a mean accuracy of 65.1%, which is 5.5%,5.8% and 8.5% higher comparing with the k means initialized Gaussian mixture model algorithm, fuzzy C means algorithm and basic Gaussian mixture model algorithm separately. It also gained a standard deviation of 11.5%, which is 5.6% lower comparing with level set algorithm. The results in these experiments proved the effectivity of the developed method in the cervix region segmentation from colposcope images.
Key wordscervix cancer screening      cervical image segmentation      k means algorithm      Gaussian mixture model     
Received: 05 May 2017     
PACS:  R318  
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Liu Jun
Yu Tingting
Shi Huijuan
Cite this article:   
Liu Jun,Yu Tingting,Shi Huijuan. Cervical Image Segmentation Based on Modified k Means Algorithm and Gaussian Mixture Model[J]. Chinese Journal of Biomedical Engineering, 2018, 37(2): 138-145.
URL:  
http://cjbme.csbme.org/EN/10.3969/j.issn.0258-8021.2018.02.002     OR     http://cjbme.csbme.org/EN/Y2018/V37/I2/138
Copyright © Editorial Board of Chinese Journal of Biomedical Engineering
Supported by:Beijing Magtech