Analysis of optimized segmentation algorithms for medical images

Abstract

There are 13.9 million new cancer cases and 8.1 million cancer deaths newlineoccurred worldwide as per the latest global survey, which makes a common newlinethreat to all. Generally, different types of malignant tumor are made up of newlinespecific types of cancer cells, including Carcinoma and Sarcoma, etc. A brain newlinetumor is one of the aggressive tumors, which causes more human death. So newlineEarly diagnosis plays an essential role in improving treatment possibilities newlineand increases the survival rate of the patients. Hence, the identification of newlinetumor is necessary for successful diagnostics. Usually, this task is performed newlinemanually by experts, which is not always apparent due to the high diversity in newlineappearance of tumor tissue, among different patients. Thus, automating tumor newlinedescription is a real challenge which has attracted the attention of several newlineresearchers in past years, which focus on the recognition of regions in newlineMagnetic Resonance Images, but yields limited accuracy. In the work, an newlineexpansive number of models and algorithms have been proposed to address newlinethe issue distinguish between normal and abnormal region. newlineMulti-Variant Graph-based energy estimation is to compute the newlinefeatures to perform the segmentation. Expectation Maximization Accuracy is newlinea combination of MVG and LAS for the better solution, which has improved newlinethe accuracy. Heuristic Algorithm with Fuzzy C-Means clustering produces newlinethe best outcome for images, which contains less noise. Enhanced Levy Flight newlinebased Firefly algorithm has been found that the performance is outstanding newlinecompared to any of its predecessor hybridized breeds. It proves in dealing newlinewith worst-case images. The results are analyzed based on accuracy and its newlinerunning time newline newline