Medical Image Segmentation via Fuzzy Connectedness-based Fuzzy C-Means Method

摘要

We present a two-stage image segmentation method combining fuzzy connected object extraction and fuzzy C-means method. First, by selecting a seed point in the object of interest and computing fuzzy connectedness value between the seed point and other elements of image, a fuzzy object is extracted and the connectivity scene of the fuzzy object can be got. Second, the original image scene and the new connectivity scene constitute a two-feature cluster space, in which we can use fuzzy C-means method to refine the segmentation result. So our method has better robustness to noise and artifacts. We put this method into practice in our medical image analyzing and processing system, and the results show that this method performs well for some medical images.