Investigations on optimization techniques based brain tumor image segmentation

Abstract

Neuroimaging plays a key role in medical field. The main requirement in medical newlinefield is to identify methods which will be able to diagnose the disease exactly. Medical newlineimage segmentation and recognition techniques have a significant task in computer newlineaided diagnosis. In analyzing the brain diseases, various regions of the brain must be newlinesegmented from Magnetic Resonance and Positron Emission Tomography images. The newlineabnormalities in brain cells are the key sources for making lesions in brain. These newlineabnormal lesions in brain lead to the establishment of tumors in brain. There are newlinedifferent types of tumor which possess dissimilar characteristics and want diverse newlinetreatments. Brain tumors are categorized into primary brain tumors and metastatic or newlinemalignant brain tumors. The primary tumors start in the brain and are persuaded to stay newlinein the brain. The metastatic or malignant tumors commence as a cancer in the body newlineother than brain and then start to spread into the brain region. Benign can be treated by newlineradiation methods and malignant lesions are cured by exact surgery by skilled newlineradiologist. An early detection of brain tumor increases survival of human by providing newlinethe correct treatment. But the accuracy of segmentation is affected by artifacts. Since newlinetraditional methods do not show considerable peak signal to noise ratio and accuracy in newlinesegmentation within less amount of time, efficient methods are required. The objective newlineof our work is to develop an optimized system which has high PSNR and achieves newlinegreater accuracy in segmentation task in less computation time. newlineThe key findings of this thesis are, three optimized efficient algorithms are newlineinvestigated for brain image segmentation. Optimization strategies were effectively newlineapplied to partition the tumor portion in the segmentation area to wrench the newlinesegmentation function. In this work both brain MR and PET images are analysed and newlineregions like cerebrospinal fluid, gray matter and white matter which are the more newlineinformative regions are segmented to study and characterize t