Re-Enhancing Fuzzy Clustering for Medical Image Segmentation

Blurred image edge contours result in inaccurate image segmentation. In particular, medical images rely heavily on high-frequency information to segment different tissue and lesion regions. To do so, this work proposes a novelty fuzzy clustering algorithm for medical image segmentation by re-enhancing high-frequency information. In it, noise distributions are precisely analyzed. The noise-smoothing schemes of single and mixed noise models are given respectively. Different pyramids for different noise models are used in the proposed algorithm, which aims to fuse high-frequency information. Experimental results indicate that the proposed algorithm outperforms its counterparts in terms of effectiveness.