Safe engineering application for anomaly identification and outlier detection in human brain MRI

Image segmentation is a process that helps to engage multiple fractions of the given input medical images for diagnostic purposes. The primary objective of segmenting the images is detecting the boundaries and the localization of the object. In the Magnetic Resonance Images (MRI) of the brain, the tumor/aneurysm portions are not envisioned due to the presence of white matter (WM), gray matter (GM) and cerebrospinal fluid (CSF), and some other noises present in it. At present, this is a huge problem for the doctors to identify the tumor exactly. In this paper, we proposed a method for safe engineering application that combines Fuzzy K-Means (FKM) clustering and Artificial Bee Colony (ABC) Optimization to produce the segmented image of the MRI brain. ABC optimization is one of a stochastic method, based on the population of bees and its behavior that helps in obtaining solutions for numerous optimization problems. Many researchers generally preferred FKM, because it is an exclusive algorithm for the clustering process. Compared to the conventional clustering methods, FKM helps to segment the tissues of brain (GM, WM & CSF) and the tumor/aneurysm region efficaciously. The proposed method is tested on both low-grade and high-grade images of Brats 2013 dataset, the segmented results and the performance measure values obtained are remarkably better compared to earlier methods and they are applied in safe engineering application.