Brain tumor segmentation approaches: Review, analysis and anticipated solutions in machine learning

Abstract

Brain tumor is one of the most rigorous diseases in the medical science. An effective and efficient analysis is always a key concern for the radiologists in the premature phase of tumor growth. At first sight of the imaging modality like in Magnetic Resonance (MR) imaging, the proper visualization of the tumor cells and its differentiation with its nearby soft tissues is somewhat difficult task. The reason for the above problem is the presence of the low illumination in imaging modalities. One of the solutions of such problem is deal by using machine learning based system diagnosis. In past various segmentation methods had been applied on brain MR imaging system to figure out the proper abnormality region from overall volume of the brain. In this paper a decade survey analysis is presented for all such approaches which are used in machine learning system for tumor segmentation. Further, the paper presents the limitations and advantages of all such approaches in machine learning based diagnosis. At last, the comparative segmentation results are discussed with certain clustering performance measures to analyse the effectiveness of each algorithm.

Brain Tumor Segmentation using SLIC Superpixels and Optimized Thresholding Algorithm

“…An automated extraction of potential tumors from brain scans is thus desirable and plausible. Much research and implementation have gone into tumor segmentation using the general image segmentation techniques which include Kmeans clustering [2], [3], Fuzzy C means clustering and watershed methods [4] [5] and artificial neural networks and machine learning techniques [6,7]. Other techniques like histogram based methods [8] and region based methods (region splitting, growing and merging) [9] have also been exploited.…”