Brain Tumor Detection in Magnetic Resonance Images (MRI)
ABSTRACT
Brain Tumor Detection in Magnetic Resonance Images (MRI) is very hard in medical diagnosis as it provides data related to anatomical structures with potential abnormal tissues, necessary for the treatment and continuous monitoring of the patients. Our first stage of research work mainly focuses on the enhancement of magnetic resonance (MR) images. In this study, to improve the performance and reduce the complexity involves in the medical image enhancement process MR images of the brain, we enhance the brain tissues into healthy tissues such as white matter, gray matter, cerebrospinal fluid (background), and tumour-infected tissues. The FSHE method has to be followed for the image enhancement process—the precise experimentation implemented in MATLAB simulation environment. From the experimental results performed on the different images, it is clear that the analysis for the image enhancement process, which seems to be is fast and accurate when compared with the other existing methodologies performed by clinical experts. Then in our second stage of research work mainly focuses on the segmentation of the tumor area from the magnetic resonance (MR) images. The segmentation is crucial to get the tumor area. Accurate measurements of tumor growth in brain diagnosis are quite tricky because of diverse shapes, sizes and appearances of tumors. For that, the ICA – LDA (independent component analysis – linear Discriminant analysis) algorithm with ARHE (adaptive region-based histogram enhancement) model implemented. The appearance of tumors on MR images varies differently due to the growth of the amount of tissue variation inside the tumor area and their diffusivity. Typically it’s challenging to segment and classify the tumor area from an irregular tumor shape. To overcome this problem, we implement an ECNN model. This model can overcome all the challenges that, when compared to that of the other existing method, our method shows the best accuracy results that is a maximum of 97.33% of the approx results.