Identification and Classification of Brain Tumor from MRI Data

Abstract

During the last decade, medical practices have undergone drastic changes newlinewith large-scale involvement of the engineering support system. newlineTraditionally, diagnosis and analysis of medical illness were solely newlinedependent on experience and prior knowledge of medical specialists newlinewhich was purely manual process, long and tedious process and newlinesometimes erroneous result may lead to any kind of disastrous impact on newlinethe subject. Nowadays various medical imaging modalities have been newlineextensively used to minimize the above complexities and enable the newlinepatients to live a long and healthy life. However, ongoing through newlineliterature survey, it has been observed that certain issues and problems are newlinestill existing which could not be dealt by earlier research in order to detect newlinethe normal and malignant brain scan accurately and efficiently. Keeping newlinein view these untouched issues, we have focused on some methodologies newlineto improve the overall performance of diagnosis, as compared with the newlinestate-of-art methods, which are briefed below. The first methodology newlinecontains hybrid feature extraction technique to classify the brain tumor newlinewhich may assists radiologists and physicians to make their decision faster newlineand accurate. The second methodology contains Hybrid Deep Neural newlineNetwork architecture which is a combination of two different Neural newlineNetworks. The third methodology contains the combination of Statistical newlinefeature extraction technique with dual classification method. The fourth newlinemethodology contains Ranklet Transformation feature extraction method newlinewith combined classification technique and the fifth methodology compares newlineArtificial Neural Network, Random Forest and Support Vector Machine newlinetechniques, utilizing statistical parameters. newlineTo begin with, a new method for feature extraction ... newline