An Efficient Fuzzy Connectivity Method for Airway Tree Segmentation Using Fuzzy C-mean Algorithm

Traditional segmentation techniques do not quite meet the challenges posed by inherently fuzzy medical images. Recently, Fuzzy connectedness, FC method is introduced to capture the fuzzy nature of medical images by combining two fuzzy features based on their affinity functions. These features are an object-feature based and a homogeneity based affinity components. The main idea of FC is to capture two inherited fuzzy nature of medical images called closeness and hanging togetherness of pixels. However, this methods suffers from two drawback; manually selection of initial seeds which is elaborative and time consuming and computational burden of searching all possible path in order to assign the strength of connectedness. To overcome the above limitations, the proposed method utilizes FCM algorithm as the first step to select fuzzy features corresponding to airway wall and airway lumen and then applied fuzzy connectedness algorithm based on those features. The proposed method performs an unsupervised automatic segmentation of airway tree as well as performing a quick search along pre-clustered pixels to assign affinities.