Brain Tumor Segmentation of MRI Images Based on K-Means and White Shark Optimizer

Due to the intricacy of the process involved in segmenting and extracting the tumor regions in magnetic resonance imaging (MRI), it is quite difficult to successfully detect the diseases. Due to its critical significance in subsequent image processing phases, this creates a requirement for a suitable clustering algorithm. In this study, a clustering method is presented that may address a variety problems for MRI brain tumor segmentation. One of the most well-known conventional clustering methods that is simple to use, quick to process, and ensures convergence is k-means (KM). Its weakness, which is defined by its susceptibility to the initial center's random initialization, is present. Because of this, the KM was optimized in this study using the white shark optimizer (WSO) to address this weak side of the KM. The KM would set the starting point for WSO's final position. Evaluation results of WSO-based KM were judged on a publicly available datasets in terms of several relevant evaluation methods and contrasted with outcomes of KM, fuzzy c-means (FCM), and other meta-heuristics when practiced to cluster the same datasets. The WSO-based KM clustering approach outperformed the standard KM and other competing algorithms-based KM.