International Journal of Artificial Intelligence and Neural Networks
Author(s) : AYAT ALROSAN , NORITA NORWAWI , WALEED ALOMOUSH , WEDIAH ISMAIL
Fuzzy clustering algorithms (FCM) have some disadvantage. The main disadvantage is the cluster centroids initialization sensitivity and trapped in local optima. This study proposed a novel clustering method by coupling artificial bee colony with fuzzy c-means (ABC-FCM) algorithm. The technique exploits the superior capabilities of ABC in searching for optimum initial cluster centers and uses these clusters as the input for FCM, thus improving the segmentation of MRI brain images. The performance of the newly developed approach was tested using two sets of MRI images: simulated brain data and real MRI images.