Examining the Pathological Portions in MR Brain Slices using Automated Map and Improved Fuzzy K-Means Clustering

Abstract

Identification of pathological structures (tissue and tumor region) in brain MR images is executed by an automated algorithm, and it requires improvement in processing time and segmentation accuracy. Oncological experts have predicaments in detecting the tumor masses that have similar resemblance with the tissue matters. An innovative amalgamation of soft computing algorithms, such as the automated map and clustering technique is presented through this paper. The Self-Organizing Map (SOM), a subsection of map technique, and the clustering process named the Improved Fuzzy K-Means clustering (IFKM) are used for the automated segmentation of MR brain structures in this paper. The segmentation outcomes of the algorithm are accurate for brain MR image analysis, and it was evaluated using Jaccard index (TC), Mean Squared Error (MSE), Dice overlap Index (DOI) and Peak Signal to Nosie Ratio (PSNR) values in this paper. TC and DOI values were delivered as 84.43%, 91.43%, respectively. The efficiency of this algorithm is compared with other traditional approaches, and it has been confirmed that is better visualization of brain structures, which will greatly assist during Oncological treatment.