Nature inspired optimization based graph cut for brain MRI segmentation

Abstract

In the emerging field of digitization and image processing, mathematicians encounter various newlinetasks. It comprises examining how to boost efficiency and precision of practical applications newlinethrough image processing. Image processing in digital format has gained enormous relevance in newlinethe academic community as well as in practice in recent years, opening new opportunities for newlinemultidisciplinary studies to address these difficulties and providing novel possibilities for the newlinepurpose of research. In image processing, image segmentation plays a vital role. Process of newlinesegmenting an image into a group of objects and backgrounds is known as image segmentation. newlineImage segmentation is useful in medical imaging for extracting features, measuring images, and newlinedisplaying them. Multiple brain related diseases need volumetric study of brain tissues, and newlinesegmentation of magnetic resonance imaging (MRI) for early and appropriate diagnosis. newlineWithin the domains of image processing, image segmentation using graph-based approaches newlinehave received a lot of attention. These techniques turn segmentation problem into graphs and newlinesolve them as a problem of graph partitioning. In graph-based segmentation, finding the newlineappropriate partitioning with the smallest cut value is critical. newlineAccording to the findings, there is need for more research into the development of an effective newlinegraph partitioning method. We have carried out research on Nature Inspired Optimization based newlineGraph Cut for Brain MRI Segmentation with an objective of achieving optimal partitioning with newlineminimum cut value. Examined and studied the graph partitioning technique for Brain MRI newlinesegmentation. Graph based segmentation approaches reported in the literature have focused only newlineon local features. We have developed hybrid technique that uses a combination of enhanced newlinenormalized cut and watershed transform to include both local and global features. The newlineBraTS2020 Dataset is used for performance evaluation of developed technique.