Study of A Novel Automatic Segmentation Algorithm for MR Spine Image
YE Wei1, TAO Jing2, CHEN Xiaoyu1, LIN Qianzao1, WEI Ling1, ZHANG Xiaobing1, CHEN Wanhong1
1.Department of Imaging, Xuzhou Medical University Affiliated Hospital of Huai’an, Huai’an Jiangsu 223002, China;
2.Department of Imaging, The Fourth People’s Hospital of Huai’an, Huai’an Jiangsu 223002, China
Abstract:Objective This paper proposed an application of fuzzy C-means clustering algorithm, which was used for the automatic
segmentation of vertebral CT image. Methods First, we pretreated the MR images using anisotropic diffusion filtering for
preprocessing. Second, we used Gaussian kernel density estimation curve to obtain the optimal initial values of clustering centers,
and then segmented the image using the fuzzy C-means clustering algorithm. Finally the morphological operation was used to extract
the spine images from the muscle background. Results The MR spine images were simulated by different segmentation algorithms.
The qualitative analysis showed that the fuzzy C-means clustering algorithm was good at preserving the edge of the image. The
quantitative evaluation results showed that the fuzzy C-means clustering algorithm obtained the larger Dice similarity coefficient
and the smaller Hausdorff distance than others methods. Conclusion The fuzzy C-means clustering algorithm is a feasible MR spine
segmentation algorithm, which is more robust and accurate than other algorithms.
叶伟1,陶晶2,陈小宇1,林千早1,魏玲1,张小兵1,陈万洪1. 一种新颖的MR脊柱图像自动分割算法研究[J]. 中国医疗设备, 2018, 33(9): 61-64.
YE Wei1, TAO Jing2, CHEN Xiaoyu1, LIN Qianzao1, WEI Ling1, ZHANG Xiaobing1, CHEN Wanhong1. Study of A Novel Automatic Segmentation Algorithm for MR Spine Image. China Medical Devices, 2018, 33(9): 61-64.