The tumoral segmentation of MRI brain images remains a challenging problem. The value in the continuous gray scale interval is searched using the threshold estimate. The optimal threshold value is searched with the aid of the ABC algorithm. To obtain an efficient fitness function for the ABC algorithm, the original image is decomposed using discrete wavelet transforms. Then, by performing a noise reduction to the approach image, a reconstructed filtered image is produced with low frequency components. The FCM algorithm is used to group the segmented image that helps identify the brain tumor.
Image segmentation plays a significant role in medical applications for extracting or detecting suspicious regions. In this paper, we propose a new methodology of image segmentation based on the artificial bee colony algorithm (ABC) to extract brain tumors from magnetic reasoning images (MRI), one of the most useful tools to diagnose and treat cases doctors. The proposed methodology comprises three phases: improvement of the original MRI image (pre-processing), segmentation using the ABC-based image processing method (processing) and brain tumor extraction (post-processing). The proposed methodology compares and analyzes in total 9 MRI images taken in different positions of a patient with methodologies based on K-means, F-Fuzzy C-means and genetic algorithms.