A Modified Improved Possibilistic c-Means Method for Computed Tomography Image Segmentation

摘要

Image segmentation of computed tomography (CT) images is very difficult because of the complexity and diversity of these images. Due to the CT images have ambiguity on gray, geometry and knowledge, fuzzy set theory is used for image segmentation of CT images. The common fuzzy segmentation methods include fuzzy c-means (FCM), possibilistic c-means (PCM) and improved possibilistic c-means (IPCM). The IPCM method overcomes the drawbacks of FCM and PCM methods, can effectively detect the issue of duplication of data and noise. However, this method has low efficiency and is difficult to deal with complex data structures while working with clusters of spherical type. To overcome the drawbacks of IPCM method, a modified IPCM (MIPCM) method is proposed in this paper. The proposed method introduces the idea of weighted samples, improves the robustness of the method by relaxing constraint on possibilistic membership, while reducing the number of iterations by modifying the formula of parameter calculation at the same time. Meanwhile, this method introduces neighborhood constraint, enhances the spatial correlation, and increases the resistance to noise. Experimental results of CT images segmentation show that, the proposed method is much more efficient than IPCM method, meanwhile the CT images segmented by the proposed method have low error rate and high signal-to-noise ratio.