Energy Noise Detection FCM for Breast Tumor Image Segmentation
Leyi Xiao, Han Qi, Chaodong Fan, Zhaoyang Ai · IEEE Access · 2020
The traditional fuzzy C@hyphenmeans (FCM) algorithm ignores the spatial information of pixels and is sensitive to noise images. To overcome this disadvantage, this paper proposes a fuzzy C@hyphenmeans clustering algorithm based on energy noise detection (ENDFCM) for mammographic image segmentation. Based on the weighted mean filtering, the algorithm employs the information of the spatial pixel points to enhance its noise resistance. By introducing the energy curve and designing the spatial distance, it constrains the objective function to increase its parameter adaptability. Then, it derives the new subordinate function to increase the relevance of its subordinacy, which is updated continuously during the clustering process. The experimental results show that the algorithm has better segmentation accuracy and robustness compared with the traditional FCM and its improved methods, and can accurately segment the noisy breast tumor images.